Comparability Issues and Adjustments in Recent Royalty Rate Litigation ABD Ltd. v. South African Revenue Service

In South Africa’s first transfer pricing litigation on the subject of arm’s length royalty rates, a South African multinational telecommunications company (Taxpayer), prevailed against the South African Revenue Service’s (SARS) contention that the royalty rate charged to its subsidiaries (1% of turnover) on a uniform basis for use of its brand intangibles was too low. In this piece, we’ll explore how comparability issues were addressed in ABD Ltd. v. SARS and how the experts may have used publicly available royalty rates data to assess the materiality of certain comparability factors.

While both parties proposed arm’s length pricing based on the “Transactional Profit split Method” (TPSM), the decision ultimately hinged on a single internal Comparable Uncontrolled Transaction (often called a CUP, or Comparable Uncontrolled Price)—a trademark license agreement between the Taxpayer and a Cyprus telecom company that was also a former subsidiary of the Taxpayer—and the testimony of the Taxpayer’s lead expert witness, Dr. Ednaldo Silva, the founder of RoyaltyStat, which was purchased by Exactera in 2023.

The CUP Method is one of the original transfer pricing methods. It entails finding sufficiently similar transactions to the tested controlled transaction with which to “benchmark” an arm’slength price. Normally analysts will attempt to find a set of sufficiently comparable transactions to calculate an arm’s length range (usually based on the interquartile range), but a single transaction can be used if it meets the strict criteria set out in the OECD Transfer Pricing Guidelines (TPG) and/or local transfer pricing legislation.

As is often the case in controversies involving CUPs, SARS contended that there were comparability differences between the uncontrolled and controlled transactions that disqualified it. The decision considered the “size of market, the extent of competition, the availability of substitute goods and services, the nature and extent of government regulation of the market,” as well as duration and exclusivity, as comparability differences.

The Taxpayer’s experts countered, and the judge agreed, that while the TPG list various factors that may affect price, not all factors have a material effect on royalty rates. In fact, the TPG refer often to the importance of materiality, including Par. 1.130:

“The facts and circumstances of the particular case will determine whether differences in economic circumstances have a material effect on price and whether reasonably accurate adjustments can be made to eliminate the effects of such differences.”

Dr. Ednaldo Silva (referred to as Dr. John) testified that based on his experience with license agreements at RoyaltyStat from 1999 to 2023, the differences raised by SARS were not material.

“[59] “Dr. John: Well, here there are several factors and I have conveyed that these factors are not relevant for determining royalty rates. They may be relevant for determining the asset value of the property but not the rate of the income.”

[60] In relation to the measurability challenge, Dr. John relied on his own resource that he keeps of royalty values for his former company known as RoyaltyStat. He describes RoyaltyStat as a company that among other things provides:

“Comparable royalty rates (in the form of a commercial online database) drawn from proprietary compilations of publicly available license or similar. (such as asset purchase) agreements within the meaning of the OECD Transfer Pricing Guidelines…”

[61] Based on his experience of this database, first set up in 1999, he said that the differences were never material in royalty agreements. Thus, what Dr John does is to give both a theoretical and an empirical (based on his own research) response to the criticism that the Cyprus CUP was not a true comparable.”

While the decision does not provide details on his analysis, using the RoyaltyStat Database of over 24,500 unique license agreements, we can explore two contractual differences that were a point of contention: exclusivity and duration. To do so, we perform a “benchmark” search for pure-play trademark/trade name license agreements, and test them using RoyaltyStat’s statistical tools, including Regression Analysis.

Search Criteria

Exclusivity Statistics

Duration StatisticsRegression Statistics[1]

[1] From www.royaltystat.com: Economic theory is concerned with relations between variables, such as output and input, price and cost, royalty rate and license contract duration. Regression is designed to estimate the parameters of the independent, x variables, on the dependent, y variable. RoyaltyStat® regression function is postulated: y(i)=α+β1×1(i)+…+β5×5(i)+random error, where the index i=1,2,…,n counts the comparable license agreements in the sample. The random errors are assumed to have a fixed standard error, estimated by the standard deviation of the residuals.

y is royalty rate [% of licensee’s net sales],
x1 is duration of the contract [measured in years],
x2 is exclusivity [0 or 1],
x3 is patent number [0 or 1],
x4 is trademark number [0 or 1]
x5 is additional payments, such as upfront fees or milestones [0 or 1], and
x6 is tiered royalty (multiple rates) [0 or 1].

See Norman Draper & Harry Smith, Applied Regression Analysis (3rd edition), Wiley, 1998, pp. 307-308 (Interaction terms involving dummy variables). https://doi.org/10.1002/9781118625590.ch14.

If we dig deeper using RoyaltyStat’s Regression Analysis, we observe that exclusivity has a t-Stat of only -0.130. Generally, a t-Stat of at least 1.96 indicates statistical significance so not only are the results of exclusivity perverse (negative), but they are also statistically insignificant.

It is often assumed that because exclusivity provides greater market power, the licensor would command a higher royalty rate. However, in practice, licensing negotiations are nuanced and other factors, such as the licensee’s margins and the strength of the brand, appear to carry more weight.

Duration

The Taxpayer’s supporting expert (referred to as Ms. Sana) testified that “duration does not have a material influence on the value of the brand and hence the royalty. This she based on her many years of encountering such agreements. But she also offered a rationale. She said a longer agreement meant less risk for the licensor and hence it might agree to a lower royalty i.e. even less than 1%.”

Both the summary statistics and regression statistics indicate that a duration of 10 years is associated with a lower royalty rate than a duration of three years in brand licensing agreements.

Takeaways

The CUP Method is recognized as the most sensitive to comparability differences, but that does not necessarily mean that comparable transactions must be “perfect.” Under OECD Guidelines, the analyst must test whether comparability factors are material, and if they are, make accurate adjustments. If adjustments cannot be made for material differences, then this may call into question whether the method is the most appropriate.

Even if we cannot find external comparables, we can zoom out and use publicly available licensing data to observe whether or not certain comparability factors are generally material and estimate reasonable adjustments using statistical analysis.

Up Next

In three ongoing/recent litigations concerning royalty rates, comparability adjustments have been an issue of debate (or in certain cases a notable lack of debate), with adjustments applied in the cases of Medtronic v. Comm’r of the IRS and PepsiCo v. Australian Tax Office, but not in the case of ABD Ltd. (a South African telecommunications multinational) v. South African Revenue Service (SARS). We’ll be tackling these cases in upcoming blogs. Stay tuned.