Introduction

¶         Rarely does a U.S. court express an opinion on the intrinsic merit of a particular methodology for calculating damages in patent infringement litigation. But in the wake of several very large patent damages awards with questionable bases in economics and the law, a series of Federal Circuit decisions has started to place greater limits on damages theories.² In Uniloc v. Microsoft, one of the most important of these decisions, the Federal Circuit took special aim at the “25% Rule,” under which 25% of the infringer’s profits is the baseline determinant for reasonable royalty damages.³ The court condemned this methodology as “fundamentally flawed” and warned that evidence based on the Rule in the future will be “inadmissible under Daubert and the Federal Rules of Evidence.”⁴ The Federal Circuit later denied a combined petition for panel rehearing and rehearing en banc.

¶         Given the Rule’s tendency to yield a high starting point for a reasonable royalty determination, a plaintiff’s damages expert may yet seek to rehabilitate it for a future case despite the stern language in Uniloc. One main proponent of the Rule claims the Federal Circuit was “misled” and that its ruling is “dangerous and unnecessary.”⁵ This advocate attempts to distinguish a valid “classic 25% Rule” from an invalid “25% Rule of Thumb,” and asserts the “classic 25% Rule” should remain an admissible methodology after all.⁶

¶         The purpose of this article is to review the economic basis for the 25% Rule, taking into account the proposed distinction between the “classic” and “rule of thumb” formulations. The analysis combines insights from Uniloc and other notable cases, as well as a review of historical licensing literature, principles of economics and accounting, empirical studies of royalty rates, and reviews of actual licenses. Conflicting applications of the “classic” 25% Rule in case law show the methodology flouts basic expectations for reliability. The “classic” Rule is incapable of rehabilitation and the bar against using it should stand without qualification.

I. Justifications for the 25% Rule

¶         The 25% Rule originated as a 5% running royalty negotiated in the 1950s by Robert Goldscheider on behalf of Philco, which, at the time, manufactured televisions and other appliances. The royalty covered a portfolio of exclusive patent rights, trade secrets, trademarks, copyrights, and other intellectual property rights.⁷ Mr. Goldscheider compared the 5% running royalty in this arrangement to the pre-tax profit rate of approximately 20% for Philco licensees and calculated that the royalty constituted about 25% of the pre-tax profits.⁸ Mr. Goldscheider first wrote about using this rate as a royalty methodology in 1971, referring to it as a “rule of thumb.”⁹ In subsequent articles, he explained the Rule’s two prongs: first assuming a reasonable royalty equal to 25% of the infringer’s pre-tax profit rate, and then “tuning” the figure up or down according to the Georgia-Pacific factors¹⁰ and a host of other considerations.¹¹

¶         The Philco case itself highlights the basic problem with the Rule. To meet the requirement of general applicability, a patent damages methodology must be able to value a single patent that is naked (i.e., not bundled with other rights) and non-exclusive. But the Philco license is far more complex. The deal involved a portfolio of patents, with valuations complicated by exclusivity and other non-patent assets. The Philco example sheds no light on what proportion of the negotiated royalty was attributable to the patents as opposed to other rights conveyed. Nor does it indicate a specific value for any individual patent in the portfolio. The only definite conclusion is that the naked, non-exclusive value of any Philco patent at issue must have been only a fraction of the 5% royalty. Ironically, in an example intended to highlight the validity of the 25% Rule, the methodology fails to provide reliable information on the value of any patent and overstates the value of the entire patent portfolio by an unknown amount.

¶         Earlier professional licensing literature described royalty scenarios analogous to assigning 25% of a licensee’s expected profit as a royalty, but only in a limited set of circumstances. For example, one observer estimated possible royalties may fall between 10% and 20% of profits when the patent is for something tangible, profitable, and economically strong. Royalties may reach up to 30% of profits when an invention is “the vital element” in a manufactured product.¹² On this spectrum, a reasonable royalty based on a 25% split would require not just a strong patent, but a “vital” patent.

¶         Another observer suggested “the licensor is not entitled to more than about twenty to twenty-five percent of the estimated profits” and a 5% running royalty on sales was “just about tops for a very fine situation.”¹³ The implication is that the 25% share of profit is close to the upper bound for a royalty and would be a greatly biased measure of the value of most patents.

¶         In a 2002 article, Mr. Goldscheider claimed broader empirical support for the presumption of a 25% profit split for patent licenses. He indicated that he reviewed royalty rates in thousands of licensing transactions from the late 1980s and onward, finding a wide range of rates. In fact, he reported royalty rates ranging from 0% of sales to 77% of sales. He then collected rough data on company profits as a percentage of sales. After excluding as much as three quarters of the profit data for various reasons, he calculated an overall median profit split of 22.6%.¹⁴ Although this procedure raised numerous economic and statistical red flags, Mr. Goldscheider asserted “the data support the Rule generally.”¹⁵ To reinforce his claim that splits in this range are representative of reasonable royalties, he also cited commentators for the propositions that “most successful licensing arrangements end with royalty levels in this range [of 1/4 to 1/3 of the licensee’s anticipated profit]”¹⁶ and that a 25/75 split is “the industry norm.”¹⁷

¶         A default royalty equal to 25% of the infringer’s profit would have major implications for litigation. Such an assumption could sway triers of fact toward awarding extremely high damages because it suggests that, “on average,” such a profit split would be an empirically validated decision. Even as a starting point, there is a substantial risk that subsequent “tuning” would fail to achieve an appropriate adjustment.

¶         A recent article presented an in-depth analysis of Mr. Goldscheider’s empirical study.¹⁸ A central problem is that his data included all sorts of intellectual property transactions that did not belong in the analysis. The reported royalties in many cases bundled patent rights with other types of intellectual property, such as copyrights, as well as different exclusivity provisions.¹⁹ The patent royalty rates were also frequently attributable to multiple patents and even large portfolios of patents.²⁰ Each of these factors indicates the data are irrelevant, misleading, or too biased for determining a reasonable royalty for a single, non-exclusive, naked patent.

¶         Indeed, the vast majority of royalty rates (over 80%) in the database used by Mr. Goldscheider were essentially bare numbers with no description of the licensed intellectual property.²¹ Conclusions based on such data are inadmissible, according to the Federal Circuit’s ruling in Lucent Technologies v. Gateway, Inc.,²² which repeatedly underscores a requirement to prove benchmark royalties are comparable to those for the patent at issue by using economically meaningful criteria. Mr. Goldscheider’s sole control for comparability was to group licenses according to rough industry categories such as “consumer goods.” The Lucent decision recognizes, however, the mere fact that another license was negotiated in the same general industry provides virtually no information on the comparability of the royalty rate.²³

¶         Nearly all of the license rates associated with patents in Mr. Goldscheider’s database would require significant adjustments to be minimally comparable for calculating damages. For example, a licensing agreement between NCT Group, Inc. and Stopnoise, Inc. set a royalty of 5%, but this rate covered rights to thirty-five U.S. patents.²⁴

¶         The royalty rates in the data, even when relevant to patent licenses, were likely biased upward. The database only included information reported in publicly available sources, the most important of which were SEC filings. However, these filings generally disclose an individual transaction only when it is large enough to be material to the company’s overall financial position. Accordingly, the database did not include a representative number of transactions involving lower-value licenses. This consideration, combined with the high incidence of exclusive, non-naked, and/or multiple patents, strongly suggests the data were biased upward relative to the universe of all patent royalties.

¶         A reasonable royalty equal to 25% of the licensee’s profits might apply to a small number of patents that involve a vital and commercially important technology. But even in such cases, the rate should be proven reasonable using independent evidence and should not merely be assumed reasonable. There is no valid basis for Mr. Goldscheider’s claim of empirical support for an average royalty on the order of 25% of profits.

II. Tuning and the 25% Rule

¶         Mr. Goldscheider has often advocated the 25% split as a “starting point” for a reasonable royalty that should be “tuned” depending on the circumstances of a given case.²⁵ Mr. Goldscheider suggests this tuning adjustment, based on the application of the Georgia-Pacific factors or other factors the damages expert deems relevant, distinguishes his “classic” 25% Rule from the inadmissible “rule of thumb” he claims was the focus in Uniloc.²⁶

¶         However, this forced distinction lacks historical support. In reality, tuning has been the norm in cases using the 25% Rule to determine a reasonable royalty—Uniloc was no exception. Mr. Goldscheider suggests that Joseph Gemini, Uniloc’s damages expert, simply calculated a “wooden” 25% split in accordance with the “rule of thumb.”²⁷ But the Federal Circuit emphasized that Mr. Gemini based his opinion on a hypothetical negotiation incorporating the Georgia-Pacific factors as modifiers to the split.²⁸ The Uniloc decision therefore considered and rejected the very tuning methodology that Mr. Goldscheider continues to advocate as the “classic” Rule.

¶         As the Uniloc decision carefully explained, Mr. Gemini testified his procedure was “‘to adjust this 25% up or down’” depending on his consideration of the Georgia-Pacific factors.²⁹ In this case, Mr. Gemini simply concluded the factors in favor of Uniloc and Microsoft respectively were equally balanced and did not change the 25% starting point.³⁰ The result was a damages claim of $565 million.³¹ Regarding the basis for assuming Microsoft would have used such a starting point in a hypothetical negotiation, Mr. Gemini testified, “It’s a widely accepted rule.”³² The Federal Circuit found Mr. Gemini’s starting point bore no relation to the facts of the case, and was therefore “arbitrary, unreliable, and irrelevant.”³³

¶         The Uniloc decision further stated “it is of no moment that the 25 percent rule of thumb is offered merely as a starting point.”³⁴ The decision reasoned that “[b]eginning from a fundamentally flawed premise . . . results in a fundamentally flawed conclusion.”³⁵ Notably, Mr. Goldscheider’s effort to rehabilitate his notion of a “classic” 25% Rule does not address any of the searching criticisms the Federal Circuit specifically directed against it.

¶         Georgia-Pacific and many other cases illustrate the use of rates from other licenses to determine a reasonable royalty for the patent in suit. Such an analysis may even be the most common patent damages methodology. In this context, the 25% Rule might be viewed charitably as a “surrogate” license to provide a benchmark royalty rate when no actual license can be identified for this purpose. But the Federal Circuit’s decision in Lucent emphasized the need for close scrutiny of “comparable” licenses used as benchmarks to avoid erroneous results.³⁶ In ResQNet.com, Inc. v. Lansa, Inc., the Federal Circuit vacated a patent damages award primarily because there was no “discernible link” between the claimed technology and the majority of the licenses considered by the plaintiff’s expert.³⁷ Moreover, the expert relied on royalty rates from licenses that conveyed rights to substantial amounts of non-patent property, including software, which undermined the usefulness of such licenses as benchmarks. Even if the 25% Rule were simply viewed as another license benchmark, the implication of both Lucent and ResQNet is that a plaintiff would have to demonstrate the relevance of such a benchmark and not simply assume it. Mr. Gemini’s justification for his use of the 25% Rule, for example, skirts this basic requirement.

¶         It is not clear, however, whether one could ever reconcile the 25% Rule with Lucent. The benchmark royalty in a comparable license can be rationally adjusted for the patent in suit only if the provisions of the license are known, but such an adjustment is not possible under the Rule. According to Lucent, one would have to ascertain the subject matter of the license agreements that supposedly justify the use of the Rule and also assess whether the patents covered by those other agreements are in any way similar to the patent in suit.³⁸ Mr. Gemini’s inability to point to a comparable royalty rate while using the 25% Rule illustrates the problem. To the author’s knowledge, the details of actual license agreements have never been presented in litigated cases to scrutinize the comparability of a royalty benchmark based on the 25% Rule. Without factual support of this nature, the 25% rate, even as a starting point, is little more than “the ipse dixit of the expert” that was rejected by the Supreme Court in Kumho Tire.³⁹

¶         The “tuning” argument also fails to justify the “classic” 25% Rule as a matter of basic logic. It presumes that making adjustments to an economically arbitrary starting point leads to a valid endpoint. But if the valid endpoint is truly known, one could just as well use another starting point, such as 10% of profits or even zero, and apply the corresponding adjustments. The starting point therefore is an irrelevant detour. The damages expert should focus directly on the ultimate conclusion and show its validity. The only certainty the 25% Rule provides is to ensure a high starting point, with no assurance that the balance of the analysis will yield an accurate damages figure.

III. Profits and the 25% Rule

¶         In addition to the problems identified in Uniloc, the 25% Rule fails to provide a clear and economically consistent definition of the total profits to be apportioned. The relevant case law shows fundamentally conflicting applications of the Rule in this regard.⁴⁰ Because the 25% Rule proffers no reasoned basis for selecting one profit basis over another, the Rule is far more arbitrary than even its critics generally recognize.

¶         There are many different profit concepts in accounting and economics. For example, gross profit is equal to revenue minus cost of goods sold without accounting for any overhead expenses. Operating profit is equal to gross profit less overheads such as selling, general, and administrative (SGA) expenses. There are still other profit measures, such as earnings before interest and taxes (EBIT) and earnings before interest, taxes, depreciation, and amortization (EBITDA). One study showed that using these alternative standard definitions of profit with the 25% Rule could swing the resulting royalty rate by a factor of four or more—an extremely high margin of error.⁴¹

¶         Proponents of the 25% Rule have never agreed on which measure of profit to use. Robert Goldscheider first used 25% of gross profits.⁴² However, he later adopted operating profits and rejected the use of gross profits as “specious.”⁴³ Russell Parr described the Rule as using gross profits, but has also stated “the only appropriate application of the rule is in terms of operating profits.”⁴⁴ Richard Razgaitis argued the 25% Rule applies to the incremental profit attributable to the patented invention.⁴⁵ Jonathan Kemmerer and Jiaqing Lu argue for the use of EBITDA.⁴⁶ William Lee explained his use of the Rule in terms of the licensee’s “anticipated” profit, without referring to a specific accounting definition.⁴⁷

¶         This methodological chaos has critical implications for determining a reasonable royalty. i4i v. Microsoft, resulting in a $200 million jury award, provides a vivid example.⁴⁸ The infringing products involved versions of Microsoft Word in Office Professional 2003 and 2007 sold to businesses.⁴⁹ The plaintiff’s expert, Michael Wagner, used the 25% Rule, characterizing it as “well-recognized” and “widely used.”⁵⁰ He defined profits for use with the Rule as 76.6% of the assumed infringing sales. This rate is evidently based on gross profits or something similar¾Microsoft had an overall gross profit of approximately 80% during the claimed damages period. Multiplying 76.6% by the claimed value of the sales, at a 25% rate, yielded a starting point of $96 per unit for the claimed royalty.⁵¹ The expert then employed a tuning procedure using the Georgia-Pacific factors.⁵² Mr. Wagner made an upward adjustment from the 25% baseline, which increased the royalty to $98 per unit, so that the total claimed damages was $200 million when applied to the infringing sales.

¶         Mr. Goldscheider, however, has written, “[T]he rule is not a split of gross profits.”⁵³ Instead, he asserted the Rule should use operating profits after fully loading all operating expenses.⁵⁴ He explained that gross profits should be adjusted by allocating common (or non-manufacturing overhead) costs to the product line to derive operating profits.⁵⁵ Operating income for Microsoft during the claimed damages period was approximately 35% of revenue. Compared to the approach based on gross profits, using operating income would cut the royalty generated by the 25% Rule by more than half, reducing the claimed damages from $200 million to under $100 million.

¶         Indeed, Mr. Wagner employed a new version of the 25% Rule in i4i that lacked any basis whatsoever in economics or the terms of actual licenses. First, he defined profit based on revenues that Microsoft never made. Further, he assumed an average price of $499 for the units in the royalty base.⁵⁶ However, this price was the list price of a non-Microsoft product called XMetaL.⁵⁷ Microsoft’s expert testified this price was two times the price at which Word 2003 was actually sold.⁵⁸ Thus, the profits Mr. Wagner used to calculate the royalty with the 25% Rule were purely imaginary.

¶         Mr. Wagner jointly wrote an article in 2010 with the co-authors of the 2002 Goldscheider study (which described the use of gross profit as “specious”).⁵⁹ Remarkably, even though the joint article contained a section titled “Proper Understanding of the [25%] Rule,” that discussion did not address the critical importance of defining a profit measure. The 25% Rule suffers from an extraordinary lack of ground rules that has allowed experts to adopt radically inconsistent damages methodologies under the guise of one common approach.

IV. The 25% Rule and the Entire Market Value Rule

¶         The legal principle known as the “entire market value” rule further diminishes the relevance of the 25% Rule for measuring patent infringement damages. The entire market value rule holds that damages may be based on the value of the entire assembly when (a) a product consists of unpatented components together with a patented component; and (b) the sale of the entire assembly depends on a patented invention embodied in it.⁶⁰ The unpatented components must function together with the patented component as if they were all components of a single assembly used to produce a desired result.⁶¹ For the entire market value rule to apply, the patentee must prove “the patent-related feature is the ‘basis for customer demand.’”⁶² If so, the royalty base may then include the value of the entire assembly and not just that of the patented component.

¶         When the entire market value rule does not apply, it would be erroneous to assume a relationship between the profits (however defined) from selling the product and the value of an invention that may form merely a component of the product. The Federal Circuit decisions in Lucent and Uniloc discussed this issue extensively. For example, the infringement in Lucent concerned a “date picker” feature in the calendar included in Microsoft Outlook. The plaintiff’s expert claimed an 8% royalty on all of Microsoft’s sales of Outlook software, which amounted to $562 million.⁶³ But the Federal Circuit held the date picker was “a tiny feature” and it was “inconceivable” that the value of the date picker was a substantial portion of the value of Outlook.⁶⁴

¶         The “classic” 25% Rule described by Mr. Goldscheider is a split of the total profits generated by the product. But as Lucent explained at length, the justification for basing a reasonable royalty on the entire volume of profits for the patent at issue must be proven using case-specific evidence.⁶⁵ The 25% Rule provides no guidance or useful information for this critical aspect of a damages analysis.

Conclusion

¶         The 25% Rule has no reasonable grounding in economics. Instead, it appears to have emerged as a negotiating tactic, particularly for strong patents outside of the typical non-exclusive and naked licensing scenario for patent damages analysis. The chief example used to illustrate the Rule—the Philco licenses from the 1950s—detracts from the Rule’s soundness as the intellectual property rights at issue were not at all comparable to those in most patent cases. The agreed-upon royalty covered a portfolio of patents, provided territorial exclusivity, and conveyed many other non-patent rights.

¶         It is a misnomer even to speak of a 25% “rule” because there is no consensus on how that rule is defined. One of the biggest questions remains the definition of the profit to be apportioned in the calculation of damages. Practitioners have used measures ranging from the infringer’s gross profit to operating income, and damages calculations involving these different measures can vary by a factor of two or more. In i4i v. Microsoft, the Rule was even applied to a share of hypothetical profits on a product Microsoft never sold.

¶         The 25% Rule essentially relies on little more than a sweeping assumption about what a royalty rate should be. It provides a high starting point for a royalty that is clearly inappropriate for many or most patents. The use of total profit for calculating a royalty further aggravates the bias because the entire market value likely applies in only a small minority of patent cases.

¶         The Rule cannot qualify as a damages methodology because even its most vigorous proponents present it merely as a “starting point” for some other calculation that is supposed to lead to a correct damages figure.⁶⁶ By structuring a damages analysis in this way, the 25% Rule imposes an unnecessary and arbitrary detour that does not create a reliable end result. The damages expert should present a coherent economic analysis grounded in actual case facts and data, where reliability is assessed using statistics, valid royalty benchmarks, and other empirical tools.

¶         A long line of cases—including Polaroid v. Kodak, TWM v. Dura, Panduit, and Georgia-Pacific—shows how reasonable royalties can be determined by fact-based analyses when no established royalties are available.⁶⁷ The 25% Rule was not used in those cases and would not have improved the accuracy of the eventual outcomes.

¶         The Uniloc court properly held the 25% Rule is arbitrary, unreliable, and inadmissible under Daubert and the Federal Rules of Evidence.⁶⁸ The Federal Circuit considered and expressly rejected a damages methodology that used the 25% Rule as a starting point to be followed by adjustments based on the Georgia-Pacific factors or other factors. Nonetheless, Mr. Goldscheider continues to advocate this approach with his notion of a “classic” 25% Rule. Nothing in his argument addresses the grievous deficiencies of the rule identified in Uniloc and discussed in this article. The “classic” 25% Rule, were it to be permitted in expert testimony, would greatly increase the potential for biased and unreliable damage awards in patent litigation.

Footnotes