December 12, 2002
What's in a Name? Perhaps Plenty if You're a Job Seeker
HAT'S in a name? Evidently plenty if you are looking for a job.
To test whether employers discriminate against black job applicants, Marianne Bertrand of the University of Chicago and Sendhil Mullainathan of M.I.T. conducted an unusual experiment. They selected 1,300 help-wanted ads from newspapers in Boston and Chicago and submitted multiple résumés from phantom job seekers. The researchers randomly assigned the first names on the résumés, choosing from one set that is particularly common among blacks and from another that is common among whites.
So Kristen and Tamika, and Brad and Tyrone, applied for jobs from the same pool of want ads and had equivalent résumés. Nine names were selected to represent each category: black women, white women, black men and white men. Last names common to the racial group were also assigned. Four résumés were typically submitted for each job opening, drawn from a reservoir of 160. Nearly 5,000 applications were submitted from mid-2001 to mid-2002. Professors Bertrand and Mullainathan kept track of which candidates were invited for job interviews.
No single employer was sent two identical résumés, and the names on the résumés were randomly assigned, so applicants with black- and white-sounding names applied for the same set of jobs with the same set of résumés.
Apart from their names, applicants had the same experience, education and skills, so employers had no reason to distinguish among them.
The results are disturbing. Applicants with white-sounding names were 50 percent more likely to be called for interviews than were those with black-sounding names. Interviews were requested for 10.1 percent of applicants with white-sounding names and only 6.7 percent of those with black-sounding names.
Within racial groups, applications with men's or women's names were equally likely to result in calls for interviews, providing little evidence of discrimination based on sex in these entry-level jobs.
There were significant differences in interview-request rates among the nine names associated with black women, but not among the names within each of the other groups.
At the low end, the interview-request rate was 2.2 percent for Aisha, 3.8 percent for Keisha and 5.4 percent for Tamika, compared with 9.1 percent for Kenya and Latonya and 10.5 percent for Ebony.
Only part of this variability reflects chance differences resulting from sampling, although the authors have not been able to find a good explanation for the wide range. Thus it is important that the names chosen for black women were not uncommon; they represent 7.1 percent of all names listed on Massachusetts birth certificates for black girls from 1974 to 1979.
The 50 percent advantage in interview requests for white-sounding names held in both Boston and Chicago, and for both men and women.
This discrepancy complements findings from earlier studies in which researchers sent a small number of matched black and white "auditors" to apply for jobs in person. Typically, though not always, the black job seekers were less likely to be invited for an interview or offered a job.
Those findings, however, were criticized because the applicants knew the intention of the study and might have behaved differently. In addition, the auditors might not have been well matched with the jobs in question; they could have been overqualified or underqualified.
Professors Bertrand and Mullainathan's study is less susceptible to these concerns. First, they used a large number of names and inanimate résumés. Second, the job openings involved administrative, sales, clerical and managerial positions, and they submitted résumés patterned after real résumés of people who were actually seeking similar jobs.
Their most alarming finding is that the likelihood of being called for an interview rises sharply with an applicant's credentials — like experience and honors — for those with white-sounding names, but much less for those with black-sounding names. A grave concern is that this phenomenon may be damping the incentives for blacks to acquire job skills, producing a self-fulfilling prophecy that perpetuates prejudice and misallocates resources.
Two main theories explain labor market discrimination. One, known as taste-based discrimination, posits that employers — or customers, co-workers or supervisors — have a preference against hiring minority applicants, even if they know they are equally productive.
The other, known as statistical discrimination, assumes that employers personally harbor no racial animus but cannot perfectly predict workers' productivity. In this case, an employer assessing an applicant would assign some weight to the average performance of the person's racial group, instead of basing the judgment solely on the individual's merits.
A difference between these models is that employers sacrifice profits to indulge in taste-based discrimination, while, in principle, statistical discrimination, if based on accurate information, can help the bottom line. Professors Bertrand and Mullainathan cannot distinguish between the models — and both may be applicable — but they suspect that their finding that employers in heavily black areas of Chicago are less likely to discriminate against black-sounding names augurs for taste-based discrimination.
Nevertheless, either rationale for discrimination is illegal and prohibited.
"That which we call a rose," Juliet said, "by any other name would smell as sweet." An organization like the Civil Rights Commission or the Equal Employment Opportunity Commission could perform a service if it routinely monitored discrimination by conducting audit studies similar to Professors Bertrand and Mullainathan's.