“Most people use statistics the way a drunkard uses a lamp post, more for support than illumination.” Mark Twain
What are the losses due to “IP Theft” from China? On a recent trip to Washington, DC, I heard the range of $300 billion to $600 billion repeated from various sources without any critical gloss. These numbers have taken on a greater legitimacy than they likely deserve, in terms of capturing the scope of US concerns, the magnitude of the loss and shaping the Trump administration’s unilateral retaliation.
The 2017 and 2013 reports from the Commission on the Theft of American Intellectual Property (the “Commission”) appear to be the origin of much of this data. The data was also referred to in the Section 301 Report (p. 8) and in a subsequent White House report “How China’s Economic Aggression Threatens the Technologies and Intellectual Property of the United States and the World” (June 2018).
In 2017, the Commission found that “Chinese theft of American IP currently costs between $225 billion and $600 billion annually.” The Commission pulled together different sources of data, including sales of counterfeit and pirated goods ($29 billion), and that “the value of software pirated in 2015 alone exceeded $52 billion worldwide.” The Commission further noted that there was “a paucity of reliable data on the economic costs of patent infringement” and that American companies were most likely the leading victims of this “IP Theft.” It estimated losses of at least 0.1% of the $18 trillion U.S. GDP.
The largest single loss contributor to the Commission’s estimate was based on data provided by create.org and later repeated by the White House, that trade secret theft cost between 1% and 3% of US GDP, and totaled between $180 billion and $540 billion. One critic (Stephen S. Roach) of these loss figures recently noted that “the figures rest on flimsy evidence derived from dubious ‘proxy modeling’ that attempts to value stolen trade secrets via nefarious activities such as narcotics trafficking, corruption, occupational fraud, and illicit financial flows. The Chinese piece of this alleged theft comes from US Customs and Border Patrol data, which reported $1.35 billion in seizures of total counterfeit and pirated goods back in 2015.” One area of overlapping concern I have with Mr. Roach is the use of Customs seizure data to justify allocating as much of 87% of global “IP Theft” to China. Seizures by US Customs of Chinese originating counterfeit and pirated goods are as high as 87% of global totals. However, this does not mean that China is the source of 87% of the world’s production of these goods, nor does it address trade secret infringement or patent infringement origination. See The 2017 Commission Report at p. 3. Misunderstanding about the utility of Customs data contributes greatly to the weaknesses of many IP infringement loss estimates.
The 2013 Commission Report noted that “it is safe to say that dollar losses from IP theft are hundreds of billions per year, which is at least in the range of total exports to Asia in 2012 (valued at $320 billion).” This report pulled together several sources, including OECD data that estimated global trade in counterfeit and pirated goods as $200 billion in 2005 (p. 25). All the studies to date, including this 2013 Report, have recognized the difficulties inherent in doing accurate loss estimations, although many have also not distanced themselves from sources, such as the OECD 2005 data which had not stood the test of time.
Remarkably, the loss data itself has been relatively consistent over approximately two decades despite different methodologies and varying definitions of what constitutes “IP Theft”. During my tenure at the US Embassy (2004-2008), the typical guestimate was $200 billion to $250 billion per year. These guestimates enjoyed wide currency in Washington. For example, Congressional Reports, such as H.R. 110-617 “Prioritizing Resources and Organization for Intellectual Property Act of 2008” stated: “[I]ncreasing intellectual property theft in the United States and globally threatens the future economic prosperity of our nation. Conservative estimates indicate that the United States economy loses between $200 and $250 billion per year, and has lost 750,000 jobs, due to intellectual property theft.” This data was typically based on counterfeit and pirated goods “compris[ing] six to nine percent of all world trade, the bulk of which violates the intellectual property rights of United States businesses and entrepreneurs.” (Id.). Six to nine percent, however, easily gets rounded up to 10 percent, as Congressman Donnelly from Indiana noted at about the same time:
“It is estimated that these [counterfeit] products comprise almost 10 percent of world trade, that they are costing American companies nearly $250 billion in revenue and an estimated 750,000 jobs.”
The number was also widely adopted by IP advocates. The U.S. Chamber of Commerce in its report What Are Counterfeiting and Piracy Costing the U.S. Economy, (2004) circulated the 200- 250 billion dollar number, as did a National Geographic film. In fact, I used the $250 billion dollar figure when I was IP Attache in Beijing (2004-2008) to urge additional support for my elevation in diplomatic rank, as it seemed rather odd that I had been tasked with a problem costing nearly 750,000 jobs and I had no staff of my own. An article in Ars Technica (2008) noted that this earlier $250 billion/750,000 job number may have its origins in a Forbes magazine article from 1993. There are also references to loss calculations of 200 billion dollars per year appear from as early as 2002 in Congressional reports. To the extent that these calculations rely upon a base estimate of “approximately” 10 percentage points of world trade being in counterfeit or pirated goods, this data point harkens back to an OECD estimate of world trade in counterfeit goods at 5% in 1998. That number was however revised downward to 1.95% in 2007, at an estimated value of $250 billion. The OECD data seems to be the origin of the $250 billion IP Theft loss figures current at that time.
Other USG studies have shown a more cautious approach. A 2010 Government Accountability Office (“GAO”) study analyzed the economic effects of counterfeit and pirated goods and found that “it was not feasible to develop our own estimates [of the total value of counterfeit or pirated goods] or attempt to quantify the economic impact of counterfeiting and piracy on the U.S. economy.” Noting the lack of data as a primary challenge to quantifying the economic impacts of counterfeiting intellectual property and goods, the GAO concluded that “neither governments nor industry were able to provide solid assessments of their respective situations.”
The U.S. International Trade Commission in a well-researched report, China: Effects of Intellectual Property Infringement and Indigenous Innovation Policies on the U.S. Economy, (2010) calculated that the theft of U.S. IP from China alone was equivalent in value to $48.2 billion in lost sales, royalties, and license fees. This estimate falls within a broad $14.2-billion to $90.5 billion range. The breadth of this range is explained by the fact that many firms were unable to calculate their losses. Of the $48.2 billion in total reported losses, approximately $36.6 billion (75.9%) was attributable to lost sales, while the remaining $11.6 billion was attributable to a combination of lost royalty and license payments as well as other unspecified loss.
The current concerns around “IP Theft” as identified in the Section 301 Report include licensing of technology, an issue that is not covered by US Customs seizure data. Any calculation of losses due to IP Theft from non-payment of royalties should include estimates of lost royalties or license fees. Most of the current calculations do not include such data. Nonetheless, as I have noted elsewhere, accurately calculating lost royalties can be especially difficult as many licensors use tax haven jurisdictions to manage patent portfolios. There may be implied licenses in product purchases form OEM suppliers, and there can be valuation challenges. However, China’s relatively small role as a purchaser of US technology and its dominant role as an exporter of high tech or information technology products does suggest that it is a significantly under-licensed (infringing) economy. For example, China has leaped from exporting only 2% of the world’s information technology products in 1996, to 33% in 2015. Yet China has purchased very little technology directly from the US over the years, and its technology payments are a very small share of total trade. There is likely a huge shortfall in unpaid royalties from Chinese manufacturers.
Many discussions around IP theft have also declined to take into account cybercrime and other security threats. According to an often cited 2013 report by the International Data Corporation (IDC), direct costs to enterprises from dealing with malware from counterfeit software were estimated to hit $114 billion in 2013 and “potential losses from data breaches” might have reached nearly $350 billion.” However, data breaches are not necessarily a form of IP infringement, as they can be undertaken for other purposes, including simply injuring the computer system of a competitor through a denial of service attack.
Apart from differences in methodology, there are different definitions of “IP Theft” that should be affecting total loss calculations. The FBI currently defines “Intellectual property theft” as “robbing people or companies of their ideas, inventions, and creative expressions—known as ‘intellectual property’—which can include everything from trade secrets and proprietary products and parts to movies, music, and software.” This definition notably would exclude any non-willful infringement, i.e., where there is no “robbing” as well as trademarks – which are not specifically enumerated.
In its 2013 Report, the Commission offered some examples of “IP Theft”, which also excluded trademark protection, and failed to discuss patent protection: “IP theft varies widely in both type and method. It ranges from more commonly known forms, such as software and music piracy, to more elaborate types, such as the use of economic espionage tactics to steal complex industrial trade secrets. Each type of IPR violation harms an economy in unique ways and brings with it a discrete set of challenges that make both deterrence and enforcement difficult.”
These approaches to “IP Theft” are different from the meanings advanced for the same term in the last decade. Victoria Espinel, who has had a long and distinguished career in IP and international IP issues, testified in Congress in 2005 when she was with USTR and spoke of “IP Theft” in terms of the fight against ‘fakes’, declining to mention patents for inventions or trade secrets. This was consistent with the focus at that time on criminal copyright and trademark focus of US government advocacy in China, including the bringing of a WTO case (DS362):
“Our companies report billions of dollars in lost revenue, irreparable harm to their brands and future sales, all of which ultimately affects U.S. workers who design and produce legitimate products forced to compete against Chinese fakes. We want and look forward to working closely with you and your staff in combating the theft of American IP in China.”
“IP Theft” of the prior decade certainly appears under-inclusive in not focusing on patent or trade secret infringement. It also fails to reflect that most IP infringement is addressed by civil remedies, where questions of willfulness are secondary to the harm being caused. Criminal remedies, while important, are relatively rare in most legal systems. This approach is consistent with the TRIPS obligation to treat IP as a “private right.” Moreover, the TRIPS Agreement itself does not require member states to criminalize patent infringement or trade secret infringement. Finally, there may be grey areas including market access barriers, investment restrictions, government procurement restrictions, informal government supported forced technology transfer, or aggressive use of antitrust laws that many would argue need to be included in the definition of “IP Theft”. Many of these would also be very difficult to quantify.
“IP Theft” is also slightly over-inclusive, as there are also certain forms of bad-faith behavior that may be sanctioned by the state and permissible under international IP rules. For example, rights holders in China face a significant burden of bad faith patent and trademark applications that entail costs of challenging and invalidating these rights, while US-based rights holders often complain about non-practicing entities and patent trolls.
Individuals who might suspect an exaggerated “IP Theft” loss estimate might be surprised to know that there are data points that have typically been omitted from these calculations. For example, US Customs data typically does not include the value of goods excluded from the US market under Section 337 exclusion orders. I am unaware of any methodology that attempted to extrapolate from US damage awards in US courts against Chinese infringers. USTR in its 301 report, was also unable to calculate the value of “forced technology transfers” in joint ventures or technology transfers. Certain rights, such as plant varieties and plant patents are typically not included in loss figures, nor are losses due to design infringements. Consequential damages (attorneys fees/court costs/losses to brand value/harm to public health or safety) are also often not included in the above calculations.
There are also factors that could reduce the loss figures that have often not been used. Assumptions about the US being the overwhelming victim of “IP Theft” are hard to substantiate. I suspect that different countries and industries bear different costs in different markets. European companies, for example, likely suffer most from trademark and design infringements of luxury goods. In the United States, over 50% of US patent applications originate with foreigners; logically this may mean that a substantial portion of the injury suffered by US companies overseas due to patent infringement may be attributable to innovations that occurred outside of the United States.
Calculating how much “IP Theft” originates from China also ignores whatever the “baseline” is for infringement in the US. Historically, for example, the greatest value of software piracy losses were in the United States, According to BSA data for 2017, China’s piracy losses were 6.8 billion, while the US was 8.6 billion. In addition to other deficiencies, US Customs data is based largely on the pro-active behavior of US rightsholders and thereby considerable selection bias. As one example, if “IP Theft” priorities were based on Custom data, apparel, watches and footwear would be the major area of US trade concern with China, as there were the three categories of goods most seized by US Customs in 2017.
“IP Theft” losses also do not necessarily reflect losses due to unredeemed WTO commitments, nor are they based solely on violation of WTO disciplines. The TRIPS Agreement, for example, does not require members to criminalize willful trade secret theft – which is likely a major contributor to the current calculations. Moreover, if the calculations were one that adhered to WTO procedures, then the various methodologies would also need to look to WTO jurisprudence in terms of calculating damages when a WTO member fails to implement a WTO decision involving IP. A good reference point might be the “Irish Music” (DS160), which the US lost, and where the US was required to pay 1.2 million euros per year as an arbitral award in the early 2000’s. Another reference is the Antigua/gambling dispute, where the island of Antigua was permitted to retaliate against US IP interests in the value of 21 million dollars per year, considerably less than a claim by Antigua of 3.44 billion dollars.
Tying tariffs to losses due to IP theft has other challenges. While it may help address a sense of national outrage, the unilateral imposition of tariffs on Chinese imports is unlikely to benefit any US victim of IP theft, nor do the tariffs themselves appear to be geared to a particular loss threshold. Instead, the tariffs are loosely based on loss estimates but appear primarily oriented to forcing China to change its behaviors.
A cynical reader looking at the data might conclude that large loss numbers are self-serving and make compelling rhetoric in the echo chamber of Washington, DC. Someone looking over the history of the data might, however, view their weaknesses as due to such factors as difficulties in collecting data, the growth of the Chinese economy and changes in infringement practices, and changing technologies including the growth of the Internet as a vehicle for content and goods delivery. The current focus on “technology” in the scope of IP Theft might be viewed as a belated recognition of how the Chinese economy has become more technology-oriented in the past decade.
In my view the statistics do serve as more than a “support” of the type referred to by Mark Twain, above. They also help to “illuminate” a deeply felt and sustained injury that is otherwise hard to calculate.
Note: The author (Mark Cohen) has contributed to many of the reports noted above, typically in a private capacity.
Corrections to the above are welcomed.