Assistant Professor Gregory Lane

“When I send emails, if I change the name of the person who I send the email from, it has a profound effect on the responses that I get.”

When the team heard this anecdote, it spurred a thought that led to a research experiment – and, ultimately, striking results that reveal surprising new dimensions of gender bias for firms to grapple with.

Customer Discrimination in the Workplace: Evidence from Online Sales” is a new study coauthored by Assistant Professor Gregory V. Lane, an Assistant Professor at the Harris School of Public Policy at the University of Chicago; Erin M. Kelley, an economist at the World Bank who joins the Harris Public Policy faculty in March; and coauthors Matthew Pecenco, the Orlando Bravo Assistant Professor in the Economics Department at Brown University, and Edward A. Rubin, an assistant professor in the Department of Economics at the University of Oregon.

The study investigates “the impact of gender-based customer discrimination on the productivity of online sales agents,” finding that “the assignment of a female-sounding name” for an online sales agent “leads to 50 percent fewer purchases.”

The authors of the study partnered with an online travel agency in sub-Saharan African whose sales agents regularly interfaced with their customers through a chat feature on the company’s website. The authors worked with the company to randomize the implied gender of the sales agents’ name that appeared to consumers through the chat. They also made sure that the salesperson was unaware of the implied gender on the name that customers saw.  Overall, the study included over 2,000 chat interactions with customers hailing from 70 countries.

Assistant Professor Erin Kelley

“In our study, customers made many fewer purchases – in the order of 50% – when they thought that they were speaking to a woman. That signals that, as a worker, your gender, or your perceived gender, can meaningfully impact the interaction and, therefore, your performance on the job,” said Kelley. “We have convincing evidence that customers are just less engaged, that their tone becomes less expressive, when they believe that they are speaking to a woman.”

The study found that consumers purchased 50% fewer total products from individuals they thought to be female, and that the total value of their purchases was 60% lower. Customers lagged in responding to female agents, and were less likely to transition from their initial inquiry into a discussion about purchasing. While they do not find evidence of harassment or differential bargaining this does not mean discrimination is not present. “In our day-to-day lives, discrimination is not always overt,” Kelley said.

Indeed, customer discrimination in this way is not illegal: customers can choose to make a purchase or not based on whatever criteria they deem appropriate, either consciously or subconsciously. Yet the implications for female employees who face this form of discrimination can be substantial, affecting their job performance compared to their male peers.

“When we just examine non-experimental data for male chat agents and female chat agents, the data shows no difference in sales, a marked difference from our findings,” said Lane. “A company looking at the sales performance would see no difference, and they would move on.” Yet this experiment demonstrates that there is a difference. This implies that women who end up in these roles may have to work harder to overcome the discrimination they face to achieve the same performance metrics as their male counterparts.

This effect holds steady across the 70 countries in the study. Using IP addresses, the researchers were able to tell what country the chat was originating from. While there was some variation in how different names were reacted to based on ethnicity and nationality, gender remained consistent.

The results of this research contribute to a growing literature that emphasizes the importance of studying indirect forms of discrimination. The authors provide causal evidence that customer discrimination lowers the measured productivity of female employees in the workplace by a meaningful margin, and in ways they cannot avoid.

What to do about it? “Industries that tie pay to productivity should be aware,” Kelley said. “If pay is tied to number of sales, for example, women would get paid a lot less in this context and that would be discriminatory. Those companies may need to think about pay schemes that aren't so tied to productivity measures that could reflect discrimination.”

Pecenco uses professional tennis as an example of an area that has evened out pay across gender. “One of the ways that they did that,” he explained, “is simply by trying to show female players on TV, put them at the same tournaments, and in so doing sensitizing fans, allowing them to become familiar with female players’ interests and personalities as they would with their male counterparts.”

Yet this is easier said than done in a decentralized market like the one in the study: “In professional tennis, you’ll note, there's one league, with great market power. That's one place where you can really see sensitizing customers work in a clear, easy way, so there is some difference. But smaller firms can potentially do it as well.”

Ultimately, norms must change, the paper argues, for there to be a long-term solution, as customers reconsider their priors regarding female employees.

The authors of the study were surprised by the magnitude of the issue. But this study is a randomized controlled trial using administrative data, leaving little room for other explanations of the observed facts. ``We were surprised, but the proof was in the pudding.”

A part of a growing literature around the topic, the research paper contains lessons for firms across countries, and further study is required to better understand the forces at play that make customers act the way they do and solutions to combat the effects of customer discrimination in the workplace.


The application for our 2024 admissions cycle is open. To begin your graduate school journey with Harris, start your application here. If you know a great future Harris student, please refer them here.