Employers and landlords discriminate by association, avoiding particular applicants and even discarding the entire pool of applicants if one applicant possesses attributes they dislike.
These are the findings of a new study by David Phillips published in the July 2019 issue of The Economic Journal. The author examines data from seven previous experiments that measure discrimination by sending fictional applications to real job and apartment listings and recording the responses.
He finds that employers and landlords not only discriminate against candidates with attributes they dislike, such as race, gender or age. They also discriminate against a second candidate applying for the same job or apartment when the first applicant lists an attribute they dislike. As the author explains, a prejudiced employer who observes an applicant named Jamal may throw out that application, but they might also throw out the entire pool if they worry that the one applicant says something about other applicants with less identifiable names.
According to the author, this means that discrimination has been under-estimated by up to 30%. This is because discrimination against an individual extends beyond that individual and previous studies under-estimate discrimination when applicants interact in this way. For example, a researcher sends applications from two fictional people, Jamal and Greg, for the same job or apartment listing. Jamal’s chances may be impacted by his name. However, if employers care about the whole pool of applicants, Greg also does worse (to a lesser extent) because he applied alongside Jamal. This means that just comparing the two applications miscalculates the true extent of the discrimination.
Detecting discrimination convincingly has proven to be difficult. For instance, if an employer has few employees who are black, could this be due to a lack of black applicants with skills fitting the job? Researchers frequently use correspondence experiments with fictional applicants to disentangle discrimination from differences in qualifications. Legal aid groups and regulatory agencies use a similar approach to identify discriminatory landlords and enforce legal prohibitions.
‘Do Comparisons of Fictional Applicants Measure Discrimination When Search Externalities are Present? Evidence from Existing Experiments’ by David C Phillips is published in the July 2019 issue of The Economic Journal
