Female and minority inventors are less likely to have their patent applications granted by the U.S. Patent and Trademark Office, according to a new study from the University of Georgia. Applications with feminine-sounding names, as opposed to androgynous names, are less likely to be approved, too.
The research, which won the Best Proceedings Paper Award at the Southeastern Academy of Legal Studies in Business annual meeting and appears in the Summer 2020 issue of the American Business Law Journal, analyzes patent grant rates as a function of race and gender.
“Innovation is a driver of economic success. If everyone who has the capacity and will to be an inventor engages in that pursuit, then society benefits,” said lead author Mike Schuster, who is an assistant professor of legal studies at the Terry College of Business. He conducted the research with Okahoma State University colleagues Evan Davis, Kourtenay Schley and Julie Ravenscraft.
“If we’re not engaging the entire population of the country in the innovative landscape, we’re leaving an awful lot on the table,” he added.
Previous studies have shown that female and non-white inventors are historically underrepresented among U.S. patent holders, and overall, patents tend to go to white males. Although the numbers seem to be improving, the percentages still remain disproportionately low. While there is a fair amount of literature on the number of patents granted to particular groups, less research has investigated the likelihood that a minority or female patent applicant will succeed in receiving a requested patent. Schuster and colleagues built from earlier work looking only at gender to present the first analysis of patent grant rates by race and gender.
“We were curious to see how demographic attributes correlated with patent grant rates and to look into potential sources of these variations,” he said.
To conduct the research, Schuster and colleagues analyzed data from nearly 4 million patent applications from 2000-2015. Since the U.S. Patent and Trademark Office doesn’t collect race and gender information, the researchers compared the applicants’ first, middle and last names to other databases to define gender and minority status. For instance, they looked at birth certificate records from the Social Security Administration to calculate the percentage likelihood that a first or middle name would be associated with a particular gender.
Someone with the name “Jessica” is highly likely to be a woman, for instance, but someone with the name “Jessie” is closer to a 50/50 split. The applicants with first and middle names that were “highly likely,” or 90% likely to be gender-specific, were classified as that gender for the purposes of the study.
Similarly, the research team used a database developed by Kostas Tzioumis at the American College of Greece, which classifies the racial aspects of first names. They again calculated the percentage likelihood that a name would be identified with a particular race. Applicants with at least a 90% likelihood of race identification were classified that way.
Overall, female and minority applicants were less likely to have their patent application awarded.
“Jessica is less likely to get a patent than Jake,” Schuster said. “We were interested in whether this bias came from the application itself or in the patent office.”
To understand this, the research team then conducted a secondary analysis by comparing grant rates for female or minority applications whose demographic information was apparent from their name versus those whose name was not race or gender identifying. Their analysis confirmed prior work showing that female applicants whose gender was obvious are less likely to secure a patent. They did not, however, make similar findings for minority inventors.
The research team compared grant rates for minority inventors with race-identifying names to those without such a name. The team used voter rolls from North Carolina, Florida and Georgia, which include self-reported racial attributes, to determine patent applicants’ demographic data. In contrast to the findings for gender, minority inventors with race-neutral or white-sounding names weren’t more likely to secure a patent.
In additional analyses, when looking at global applications across the world, applicants with names associated with racial minorities in the United States were less likely to succeed, as well as anyone who filed an application from outside of the country. When looking at only domestic applications, black and Hispanic applicants had a negative bias, but Asian applicants had a slight positive bias.
In a final analysis, the team looked for an additional negative bias for those who identified as both a minority and a woman. There wasn’t an added bias across the board, though they did see some interactions between race and gender, which will require further study.
The research team is interested in the mechanisms behind the discrepancies, including potential unconscious or implicit biases that may distort evaluations and decisions.
“We don’t have definite evidence that implicit biases are at work here, but from the gender side, we do know that there is some bias being introduced,” Schuster said. “The good thing is that, if implicit biases are influencing patent examination, they can be mitigated through training and thinking about how we make decisions.”
Future studies may look at other areas of intellectual property, such as trademark filings, as well as the gender and race of the patent examiners and the patent attorneys who handle applications.
“It’s possible we have children who could be great inventors, but they’re not getting exposed to the idea because they don’t have role models who look like them,” Schuster said. “This perpetuates societal problems and isn’t optimal from an economic welfare perspective or a social justice perspective.”