Discrimination against foreigners in the patent system

Gaétan de Rassenfosse
7 min readOct 26, 2020


A recent stream of the academic literature suggests that there might be discrimination against foreigners in the patent system. Firms of foreign origin may be treated unfavorably compared to local firms when filing for patent protection. Estimates suggest that foreign firms might be as much as ten percentage points less likely to have their patents granted than locals. This article takes stock of the literature. It explains what we know — and don’t know yet — about the issue. It targets intellectual property (IP) professionals, policymakers, graduate students, and anyone with a keen interest in IP, innovation, or technology protectionism.

The national treatment principle

Using the patent system to discriminate against foreigners is as old as the patent system itself. The Patent Act of 1836 set application fees of $30 for U.S. applicants, $500 for any “subject of the King of Great Britain,” and $300 for all other persons. Meanwhile, in the German state of Württemberg, the Trade Act of 1828 introduced application fees that ranged from 5 to 25 Gulden, with state residents paying the lowest fees.

The rationale for discriminating against foreigners is rather simple. As the late economist Suzanne Scotchmer explains:

Governments typically want the weakest possible protection for foreign firms in their domestic markets to facilitate free-riding on foreign technologies.

A patent is a jurisdictional right; it protects an invention from imitation only in the country that issues it. Thus, anyone can freely use a technology in countries in which it is not patented. Discriminating against foreigners is particularly attractive for governments in developing countries. Lack of patent protection provides a chance for local companies to learn from and build on advanced foreign technologies freely — by reading the patent literature and implementing the inventions in their own products or manufacturing processes.

But the learning argument is not as strong as it appears. Learning from patent documents alone is notoriously difficult. Further, learning also occurs when foreign firms deploy technologies in foreign markets and directly expose locals to these inventions. Foreign firms may be reluctant to deploy technologies in markets where they lack patent protection — further reducing learning opportunities for locals. What is more, should all countries discriminate against foreigners, innovative firms would only recoup their R&D investments in their local markets. This practice would weaken R&D incentives, especially for firms with a small local market and expensive-to-produce technologies requiring a global market.

To ensure an even playing field, governments from leading nations signed the Paris Convention for the Protection of Industrial Property in 1883. Article 2 establishes the national treatment principle, imposing equal rights between foreigners and locals. The 1995 TRIPS Agreement, administered by the World Trade Organization, further reaffirmed the national treatment principle.

Discriminating against foreigners, therefore, violates these international treaties.

Evidence of discrimination in the modern era

Photo by CHUTTERSNAP on Unsplash

We could be forgiven to think that the practice of discrimination belongs to the past. But in light of rising trade tensions and concerns about technology protectionism, this issue is coming to the fore in the academic literature. What do recent academic studies tell us?

Early studies have looked at differences in patent pendency rates (that is, the time patent applications are kept pending) and patent grant rates between foreigners and locals. For example, Kotabe (1992) showed that the Japan Patent Office (JPO) seems to discriminate against foreign applicants with longer pendency periods than for domestic applicants, whereas Western patent offices appear to discriminate against foreign applicants with lower patent grant rates than for domestic applicants.

However, such evidence does not suffice to establish discrimination. Many reasons could explain differences in grant lag or grant rates between foreigners and locals. One possible explanation could be that foreigners may take more time to respond to the patent office’s communications, leading to longer grant lags. Regarding the grant probability, there might be intrinsic differences in invention quality between foreigners and locals. Correlation is not causation, and establishing the causal impact of being a foreigner on the outcome of the patent prosecution process is key to ‘prove’ discrimination.

The economic literature offers many studies of discrimination. Bertrand and Mullainathan (2004) have proposed a rather neat way of establishing racial discrimination in the labor market. They sent fictitious resumes to help-wanted ads in U.S. newspapers. To manipulate the perceived race of the job applicant, the authors randomly assigned resumes to African-American- or White-sounding names. They found that White names received 50 percent more callbacks for interviews. Adopting such an approach to studying discrimination in the patent system is not feasible, however. One cannot just send the same fictitious patent applications several times but simply changing the applicant’s origin; patent applications are only granted to inventions that are new to the world, and duplicate inventions would be quickly spotted and rejected on the ground of novelty.

Webster and colleagues (2014) have proposed an interesting test of discrimination. They have used a matched sample of patent applications granted by the U.S. Patent and Trademark Office (USPTO) and filed at the European Patent Office (EPO) and the Japan Patent Office (JPO). In other words, they observe patent applications for the same invention in three jurisdictions (so-called ‘twin patents’). They found that Japanese firms were less likely than European firms to have their patent applications granted in Europe (and the reverse was true at the JPO). In contrast to the study by Kotabe, the authors compare the fate of the same inventions across different jurisdictions, thereby alleviating concerns about systematic differences in invention quality between foreigners and locals. de Rassenfosse and colleagues (2019) have exploited a similar method to establish discrimination using more recent data and more patent offices.

These two papers hint at the fact that there might be discrimination in the patent system.

However, this method has its limits. Although it compares the same invention by the same applicant in different jurisdictions, there might still be systematic differences across jurisdictions. For example, local firms may hire more qualified patent attorneys or know best how to navigate their national patent system’s intricacies.

de Rassenfosse and Raiteri (2020) have sought to address these concerns. They also exploit twin patents but in a statistical framework that enables them to control for a large number of variables that could affect the grant rate, such as the ‘quality’ of the patent attorney firm. They focus on discrimination at CNIPA, the Chinese patent office. Interestingly, as they start controlling for more and more variables, traces of foreign discrimination vanish. They vanish for all but for one type of patent application: those in technological areas that the Chinese government considers of strategic importance. They also find that these applications are subject to more amendments during the examination phase.

This finding led the authors to conclude that, at least in China, the patent system might serve technology protectionism.

The story does not end here

The literature discussed so far has considered ‘output’ measures, such as the grant outcome or the length of pendency. More recent research has focused on process-based indicators in an attempt to understand the root cause of discrimination. Petit and colleagues (forthcoming ECARES working paper) track several metrics related to the intensity of the search performed by patent examiners at the USPTO, EPO, and JPO. Although they find traces of discrimination using output measures, they find overall no convincing evidence that discrimination occurred during the patent prosecution process. The authors conclude that offices do not seem to discriminate against foreigners — it would seem that applicants themselves may be less prone to have their patents granted ‘at all costs’ in foreign jurisdictions.

de Rassenfosse and Hosseini (2020) also look at the examination process. They focus on the USPTO and observe that U.S. applicants put a higher level of effort in the prosecution process than foreign applicants. Overall, locals interact more with examiners than foreign applicants, both for non-granted applications and for granted patents — in other words, they push harder for a grant. Accounting for the level of effort put in by applicants significantly reduces the apparent discrimination but does not eliminate it, however.

The authors make the distinction between ‘intentional’ and ‘unintentional’ discrimination. Intentional discrimination relates to disparate treatment of a specific group of applicants. In contrast, unintentional discrimination arises when policies, practices, and rules have disparate impacts on a particular group of applicants. They show that bias against foreigners results from unintentional discrimination, but they cannot entirely rule out intentional discrimination. The authors also use the term ‘apparent’ discrimination. Lower grant rates suggest discriminatory treatment. However, it is not until we have pinned down the exact causes of this apparent discrimination that we can confidently conclude that applicants are being treated unfairly.

Collectively, these studies have only scratched the surface of the discriminatory treatment of foreigners.

We need to dig more into the data to understand the root causes of the apparent discrimination. Scholars have proposed new statistical methods and have looked at different points in the patent prosecution process. We need more statistical methods (or ‘identification strategies’ in economists’ jargon) and more metrics to identify the channels through which discrimination might occur. We also need to realize that not all offices may be discriminating against foreigners. The literature has focused primarily on the USPTO and the EPO because these offices provide high-quality data. But the opaquer offices might also be those discriminating the most. Furthermore, we should look beyond the patent prosecution process. Invalidation proceedings and litigations are natural places to look next, and some work in this direction has already emerged.

Edits since the first release:
30.10.2020: Added a link to Love et al. (2016) on litigations in the conclusion.



Gaétan de Rassenfosse

I am a Professor of Innovation Policy at Ecole polytechnique fédérale de Lausanne.