“We provide the first causal evidence that Black Americans face discrimination when trying to build a job network,” says Wladislaw Mill from the EPoS Economic Research Center.
Most jobs in the U.S. are found using information and referrals provided through informal networks. The EPoS researchers studied the acceptance rate of Black and White profiles as contacts and the answers to requests for job-related information based on 20,000 real LinkedIn users. “We document that while discrimination is widespread in general, it is surprisingly highest for young and female users,“ Mill says.
No discrimination once users are part of the network
Once black profiles have overcome the initial barrier and are part of a professional network, there is hardly any evidence of discrimination. This result was also surprising to more than 250 experts in labor economics and discrimination surveyed. The specialists had expected discrimination to continue, e.g., when users were being asked for job-related advice.
A.I. generated pictures used for professional profiles
For the field experiment, the researchers used A.I. generated profile pictures to vary aspects of race inherently assigned by birth like skin tone. Importantly, the pictures of Black and White individuals were otherwise rated as highly comparable. The researchers then created more than 400 fictitious LinkedIn profiles of young, male professionals and sent out contact requests to 20,000 Americans.
Black profiles have 13 percent smaller networks
“Our findings reveal that Black profiles’ networks are 13 percent smaller than those of their White counterparts, which is substantial,” Mill says. “We find discriminatory behavior across diverse social groups. Black users also discriminate against requests from Black profiles, but to a lesser extent than non-Black individuals.” In the second stage, job-related information was requested from the network. Overall, the results show close to no discrimination at this stage.
“Discrimination mostly takes place when job networks are being built up. This insight may help combat inequalities in labor market outcomes,” Mill says. “Our results specifically suggest that improving networking opportunities for Black individuals, e.g., through mentorship programs could be an effective measure to tackle discrimination.”