Royalty rates to assess the inter-group transfer of intangible assets (eg, intangibles or intellectual property) are the subject of persistent tax controversy in transfer pricing, including multibillion-dollar litigations involving Coca-Cola and Medtronic. In both cases, the US Internal Revenue Service (IRS) asserted that no comparable third-party licence agreements existed and used the ‘return on tangible assets’ of allegedly comparable companies to increase the royalty rates charged to foreign affiliates to figures that appear to be statistical outliers.
This chapter examines royalty rates for pricing the transfer of intangibles among ‘associated enterprises’ under the Organisation for Economic Cooperation and Development’s (OECD) Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations 2017. Associated enterprises are affiliates of the same ‘controlled group’, and intangibles are classified into ‘trade’ (ie, patents, know-how and trade secrets) or marketing intangibles (ie, trademarks). The transfer of intangibles can be via asset sales or asset licences. This chapter discusses intangible assets licences exchanged for a single (not tiered) royalty rate. It focuses on non-alcoholic beverages (ie, the soft-drinks industry) and comments on the recent Coca-Cola v Commission (155 TC No 10, 18 November 2020) tax litigation. Coca-Cola’s US operating profits were increased for transfer pricing purposes by more than $8 billion during 2007, 2008 and 2009.
The OECD member countries have agreed that the ‘arm’s length principle’ should be used to determine transfer prices for income tax purposes. In the arm’s length principle, “transactions should be valued as if they had been carried out between unrelated parties, each acting in his own best interest”. The OECD transfer pricing guidelines provide three methods to determine royalty rates:
- the comparable uncontrolled price method (ie, comparable royalty rates);
- the combined ‘net’ operating profits split method; and
- the discounted present value method if the parties cannot find comparable licence agreements.
Applying the comparable uncontrolled price method is challenging because the parties (the licensor or related licensee) must establish similar ‘net’ operating profits potential between the tested licence agreement and each comparable. In addition, for each of the next five years the associated licensor or licensee must test whether the ex-post (actual) net operating profits are within a 20% margin of error of the ex-ante net operating profits forecasted.
Comparable royalty rates without establishing similar cumulative ‘net’ operating profits
In Coca-Cola v Commissioner (2020), the IRS did not apply the comparable uncontrolled price method, which is regarded as the most reliable method. No reason was given in the cited court opinion, except for the unsupported claim that the Coca-Cola intangibles licensed to foreign affiliates were ‘unique’ and that by implication comparables could not be found. In fact, the Coca-Cola licensed intangibles have substitutes because the controlled group is an oligopoly characterised by ‘imperfect competition’ in which operating profits are a positive function of market share.
The contending parties (Coca-Cola and the IRS) should have used royalty rates for non-alcoholic beverages to at least test the reasonableness of royalty rates asserted by the IRS and held by the tax court, varying from −40.6% to +54% of the affiliate licensee’s revenue. A method that produces negative royalty rates should give pause because it points to misguided application. Data from external licence agreements demonstrates that royalty rates are fixed during the term of the licence or are tiered based on certain milestones. Royalty rates from the large corpus of uncontrolled licence agreements do not vary from one year to the next, as asserted by the IRS in Coca-Cola and Medtronic.
The IRS determined royalty rates by the residual of the actual return of tangible assets, defined as the ratio of the reported net operating profits divided by the tangible assets of the affiliated licensee, above the prescribed ‘contract manufacturing’ return of tangible assets. Contract manufacturing is a legal fiction asserting that the entity has limited business functions, limited risks and does not own intangibles. Limited risks are difficult to sustain considering Kalecki’s ‘principle of increasing risk’ as a direct function of the level of investments, including payroll and the acquisition of tangible assets such as property and equipment. The IRS defined ‘tangible assets’ as the annual sum of accounts receivable, inventory and property and equipment reported on the balance sheet of the affiliated licensees and the supposed comparables. The IRS found comparable ‘contract manufacturers’ from a mixed group of developed and developing countries. Since the wage share in developing countries is lower than in developed countries, the ‘return on tangible assets’ cannot be equal, being higher in the country with the lower wage share. The IRS royalty rates based on the residual net profits of the licensee attributed to licensed-in intangibles were determined by the simple formulas:
- royalty (t) = (ρ – 0.18) tangible assets (t); and
- royalty rate (t) = (ρ – 0.18) (tangible assets (t) / revenue (t)).
Where 18% was the prescribed return of tangible assets of the ‘comparable contract manufacturers’ and t = 1 to T is the audit year index. Rho (ρ) is the reported return of tangible assets of the affiliated licensee.
The anomalous IRS formula determines royalty rates by the ‘tangible assets turnover’, meaning the ratio of tangible assets to revenue. The IRS’ notion that tangible assets determine royalty rates is capricious because this misconception has no known theoretical (economics) basis. As stated, the IRS’ range of the residual royalty rates varied from a negative 40.6% in Egypt to a positive 54% in Chile.
“Unto a forest of distrust and fear, a darksome place and dangerous to pass” (Hamlet, III.x1.13-19) – reading the IRS litigations like a revenge play – but unlike Aeschylus’ Orestes or Shakespeare’s Hamlet, the IRS did not pause to introduce “a break in the circle”, survey the relevant facts and circumstances, reflect the anomalous royalty rates or defer action.
The same versus comparable intangibles
The US transfer pricing regulations under Section 1.482 of the Tax Code are the progenitor of the OECD guidelines. The conditions for the application of the comparable uncontrolled price method for the inter-group transfer of intangibles are more stringent than in the pluralist OECD guidelines. A stringent condition is the obligatory application of the ‘commensurate with income’ principle, which was enacted in the tax code by the Tax Reform Act 1986. The post-tax reform Section 1.482 regulations installed the ‘commensurate with income’ principle (controlled royalties must be commensurate with the revenue of the licensed intangibles) by requiring that comparable licences must have ‘similar profit potential’ as the tested licence agreement established between affiliated members of the same controlled group.
The United States transfer pricing regulations contemplate two ways of establishing similar profit potential. One simple way is if the same intangibles licence exists; for example, if the same intangibles are licensed to controlled and uncontrolled parties under similar circumstances. This is infrequent rara avis. Another complex way is if comparable licences of intangibles exist; for example, if comparable external licence agreements exist in a database such as RoyaltyStat. Comparable intangibles are determined by considering multiple qualitative factors, including exclusivity and duration of the licence (see US Treasury Regulations, Section 1.482-4(c)(2)(iii)(B) – factors to be considered in determining comparability).
In Coca-Cola v Commissioner, the court recognised the absence of the same intangibles to serve as one or more ‘comparable uncontrolled transactions’ but did not consider comparable intangibles. Hence, the arbitrary IRS formula persuaded an innocent court. But are negative royalty rates plausible, or, at the other extreme, are royalty rates that capture about 50% of the licensee’s revenue plausible? This chapter will demonstrate that royalty rates in the non-alcoholic beverage industry exceeding 9% are statistical outliers, rare events, like a Black Swan. Negative royalty rates are perverse, worthy only of dismissal.
Industry royalty rates as a test of reasonableness
RoyaltyStat contains a rich corpus of more than 22,000 unique licence agreements from which more than 800 are in the non-alcoholic beverages industry. To test the reasonableness of the IRS’ wild (unconvincing) range of royalty rates, I used the following canonical search filters (licence agreement numbers are in parentheses):
- industry: food and non-alcoholic beverages (816 licence agreements);
- sub-industry: beverages (385);
- royalty base: net sales (320);
- related parties: no (exclude them, 293); and
- tiered royalty: no (also exclude licence agreements with multiple royalty rates, 230).
According to the statistical law of large numbers, a sample of 230 observations is a large selection from which stable parameters (mean and standard deviation) can be expected. The quartiles of the 230 non-alcoholic beverages royalty rates vary from Q1 = 4% to Q3 = 6%, and the three measures of centre are the median = 5%, average = 5.934%, and Tukey’s trimean = 5%. The statistical distribution is skewed because the average royalty rate is greater than the median and the standard deviation = 4.885%.
The OECD and the US transfer pricing rules specify several methods to determine arm’s length royalty rates to price the inter-group transfers of intangibles. But tax administrations and corporations with foreign affiliates involving intangibles must avoid a priori assumptions when selecting the best (or most appropriate) method to determine arm’s-length transfer prices and they must pause when the selected method produces outliers or unreasonable conclusions. Tests of reasonableness are important to separate reliable from unreliable profit ratios (eg, the return of tangible assets) or unusual royalty rates observed on the long tails of large data distributions.
The IRS’ method of determining royalty rates based on the ‘return to tangible assets’ from contentious comparable companies produced both negative royalty rates that defy economic theory and ‘super’ royalty rates that are statistical outliers when tested against a large sample of third-party licensing transactions. Like Orestes, the IRS should step back from its arbitrary acts, which are not ordained by legislative logs, and begin to raise matters of economics and statistical principles. Disregarding the dynamic corpus of external licence agreements in existence, like excluding relevant variables in econometrics, leads to biased decisions and bias has no support in economics or statistical principles.