1 Transfer Pricing, Transfer Pricing Methods, and the Usage of Benchmarks
Transfer pricing management requires arm’s length information. Such information is necessarily deployed for ex-ante price-setting decision-making and the ex-post assessment of transfer pricing outcomes of transaction types, profit centers, or legal-entity units. The arm’s length information is built on the notion that there is an opportunity to observe third-party comparables, i.e. to identify respective data on transactions between third parties such as prices, margins, goods, services, rights, guarantees, or data on third-party units like companies or profit center units of such companies. Following this notion of the arm’s length principle, such third-party comparable shall be deployed to substantiate or establish the arm’s length nature of the related-party transaction.
To marry the related-party situation with the third-party arm’s length information, so-called OECD Transfer Pricing Methods (OECD TPM) have been defined by the OECD Transfer Pricing Guidelines (OECD TPG) since 1979 and further developed up to the current version of 2017. The OECD TPM are the models of the arm’s length test, applicable to a given fact pattern. The most prominent test models are CUP, Resale minus, Cost Plus, TNMM, Residual Profit Split, Profit Split, the like. Details on the OECD TPM will be elaborated elsewhere.
The idea of arm’s length test models along the OECD TPG is reflected by many national tax provisions and practices which obey the OECD TPG in line with the respective double tax treaty regime. Another set of tax jurisdictions may not follow the arm’s length principle as defined by the OECD, but rather the logic of the UN Transfer Pricing Manual which is primarily directed to so-called “developing countries”. And last but not least, a further set of countries deviates from the model of comparing the transfer pricing fact pattern with third-party comparable while resorting to safe harbor rules and other pre-defined sets of information to be used for testing the fair share of profitability allocated into the jurisdiction.
2 Structuring Arm’s Length Information
Arm’s length information can be structured along with the following list of indicators. The order shall indicate the decreasing relevance in practical transfer pricing cases which, most often, are determined to deliver a so-called “transfer pricing documentation” including the arm’s length test:
a) Profitability Ratios: Margins or markups
b) License Ratios
c) Interest Ratios
d) Price data
e) Cost Ratios
f) Relative figures other than financial ratios
g) Absolute values related to an observation
h) Others
2.1 Markup and Margin
With reference to a), profitability ratios are relative figures either as a margin or a markup. A margin should be understood as “an element out of an amount”, whereas a markup can be described as “an element on top of an amount”. Our default illustration of figures is as follows:
§ A 10% profitability margin might be computed, using standardized figures, by 10 divided by 100. The markup of this numerical example is 10 divided by 90 (or ‘10 on top of 90’) which equals 11,11%. If 10 is operating profit and 100 is operating revenues, such margin is called “operating margin” (OM%) and, on the level of the legal entity, the indicator is called “EBIT%”, i.e. the operating margin of the legal entity.
§ A 10% profitability markup is 10 divided by 100 (or ‘10 on top of 100’). The value of the margin in this numerical example is 9,09% (10/110), i.e. ≠ 10,00%.
From this definition, it should become clear how important the understanding of the computation of such relative figures is for transfer pricing assessment. Often, in practice, such relative figures are used without having clarified whether it is a margin or markup and whether the figure is a gross or net margin (or markup).
2.2 Gross or Net
Note that, from an assessment point of view, the net margin (or net markup) directly represents the basis of profit taxation while the gross margin (or gross markup) entails overhead costs which will have to be assumed, or covered, by this gross value.
Similarly, ratios of b), c), e), and f) are relative figures with a numerator and denominator. In the case of b), c), and usually e), the numerator of the equation is not part of the amount of the denominator, hence the ratio expresses a markup.
Also, other financial ratios (f) could be any kind of relationship between values within the P&L sheet OR between the P&L sheet and the balance sheet (e.g. the ratio return on investment, ROI%).
2.3 Price Data and other “absolute values”
Data of d), price data, in practice, are considered representing “absolute figures”, i.e. an amount. However, such values actually only add up sense to the arm’s length statement if such price data is related to other price data. A third-party price data without its context (number of pieces, quality, contractual setting, etc.) does not tell much on the arm’s length character of the transfer price.
Price data usually comes as monetary unit values such as Euro, Dollar, Yen, and the like.
Figures of g), absolute values related to observation, could be, among others, numbers of employees per sales amount, numbers of pieces per hour, and the like. In practice, such figures might be used additionally to other arm’s length indicators, however, they usually do not suffice the arm’s length test for tax purposes.
3 GTP® Benchmarks: Standardized sets of Arm’s Length Information and Interquartile Statistics
Over more than a decade, the GTP® TEAM made the observation that the so-called interquartile statistics of comparable companies, identified for arm’s length tests using profitability ratios, are pretty similar across countries, industries, and functions. Following observations can be sketched, subject to the condition that the search of comparables is of random nature and the exclusion of companies being treated not comparable is not related to its profitability but other features. The interquartile statistics can be described as follows, referring to the operating margin EBIT% [Earnings before Interest and Tax divided by Operating Revenues] as the most commonly used profitability indicator deployed in transfer pricing arm’s length analysis:
3.1 The interquartile range between Q1 and Q3 of the sample does not lay fully outside the range between 0% and 10% but always touches this default range. In some cases, the Q1-percentile might fall below the zero percent threshold, and/or the Q3-percentile could be above the ten percent level. Yet, in 15 years of conducting benchmark studies, we’ve never observed interquartile statistics with an interquartile range being fully, or by far, outside this default range of one-digit percentage range (i.e. between 0% to 10%).
3.2 As the size of the sample, i.e. the number of observations, falls below a dozen of observations, the interquartile statistics becomes “instable” in the meaning that the average mean and the median might significantly deviate and, more importantly, the interquartile range (range between Q1 and Q3) might significantly differentiate while eliminating one observation more from this sample. The more the sample’s observations are unequally distributed, the earlier this “instability” of the interquartile ranges occurs.
3.3 No stability and, hence, no statistical significance should be expected in samples of six observations or less. Such interquartile statistics are worthless for assessing transfer pricing patterns on the arm’s length character.
3.4 If statistics should play its role in the transfer pricing analysis, the sample size should be larger than 20 observations. We prefer sample sizes of 50 observations and more. In such a case, the interquartile statistics clearly provide the stabile information for the arm’s length statement.
3.5 The so-called “internet search” is considered by the GTP® TEAM as a waste of time resources spent for such studies, and this is for several reasons:
The time gap between the P&L data retrieved from the database and the internet-driven search for information of “comparability” on a given company could be up to five years and more in some instances. There is hardly found evidence on which dataset is more accurate: the internet-based data or the data from the database?
The internet-based dataset could be distorted for reasons that the internet usually is deployed for marketing purposes, while data of the database used for the benchmarking usually is based on published financial statements which, themselves, were audited for by the accounts and the financial statement of the company.
If the size of the sample remains about 20 observations, the interquartile range will keep similar – no matter whether observations were eliminated from this sample due to the “internet search” or not.
3.6 Benchmarking should become more standardized in order to keep this part of the transfer pricing system a valuable piece of information for the arm’s length assessment of the tested party’s profitability or the price setting approach of the related parties involved.
3.7 It raises astonishments if, still today, other experts claim the “non-comparability” of a set of companies without defining the specific features of comparability. Comparability usually is defined by certain factors as defined by the OECD TPG. When raising such claims on comparability, usually no specifications are provided for such as “what search variable” exactly should be picked out of the many thousand options and, if such search variable is defined, what parameter value should be applied to that variable (discrete, continuous). As such definitions are not provided for, as for example by the OECD or by the local authority (issuing regulations or administrative principles), the claim for comparability appears fuzzy and arbitrarily.
For information, visit www.GlobalTransferPricing.com