Evidence of ethnic discrimination in the Swedish labor market using experimental data

Abstract

We present evidence of ethnic discrimination in the recruitment process by sending fictitious applications to real job openings. Applications with identical skills were randomly assigned Middle Eastern- or Swedish-sounding names and applications with a Swedish name receive fifty percent more callbacks for an interview.

We extend previous analyses by adding register and interview information on firms/recruiters to the experimental data. We find that male recruiters and workplaces with fewer than twenty employees less often call applications with a Middle Eastern name for an interview.

Introduction

There are reasons to expect that discriminatory behavior exists among Swedish employers. Longitudinal attitude surveys of the general public show evidence of negative attitudes toward immigrants in general and surveys among potentially discriminated groups also point in this direction (see Lange, 2000, FSI, 2004).¹ These studies indicate that discrimination is worst against individuals with a Middle Eastern background. Also, unemployment rates for immigrants born in the Middle East are found to be several times higher than for natives, indicating that ethnic discrimination exists in the recruitment process. Whether this is true is the main question to be answered in this study.

However, ethnic discrimination is just one possible explanation for native-immigrant differences in finding jobs. Another is (to the researcher) unobserved productive characteristics. Since researchers can seldom account for all differences in productive characteristics between groups it is a difficult task to empirically identify the extent of ethnic discrimination in the labor market.

To circumvent this difficulty researchers have relied on using field experiments specifically designed to test for discrimination in recruitment.² The correspondence testing methodology implies that the researcher sends two equal applications to advertised job openings with the only difference being the name of the applicant; one with a native-sounding name and the other a foreign-sounding name. Hence, this methodology ensures that productive characteristics on the supply side are held constant. Ethnic discrimination is quantified by the difference in the number of callbacks for interview between the two ethnic groups.

Contrary to most previous studies using this methodology we proceed with collecting company and, especially, recruiter information through official registers and by conducting interviews with the recruiters. Hence, while some studies use variation in application information to analyze how discrimination varies across application attributes (see Bertrand and Mullainathan, 2004), we use the same application attributes for both applicants to analyze which firm/workplace and recruiter characteristics that are correlated with the ethnic difference in the callback rate to interview.³

Our experimental data was collected between May 2005 and February 2006 by sending two applications; one with a Swedish-sounding male name and one with a Middle Eastern-sounding male name, to job openings in twelve different occupations in the Stockholm and Gothenburg labor market areas. In total we replied to 1552 job ads posted at the home-page of the Swedish Employment Agency. It was found that the callback rate of applications with a Swedish-sounding male name was fifty percent higher than for the ones with a Middle Eastern-sounding name. Interestingly, despite the fact that low level occupations have the largest share of immigrant employees we found that the relative callback rate for Swedish-sounding named applications is higher in lower level occupations than in higher level occupations. Further, regression analysis shows that the ethnic difference in callbacks is related to the sex of the recruiter and to the number of employees at the workplace.

The remainder of this paper is outlined as follows. Section 2 presents the methodology used and the design of the experiment. Section 3 describes the collected data. In Section 4 an analysis is performed using register and interview data on which workplace and recruiter characteristics influence the ethnic difference in the probability of being called for an interview. Section 5 concludes.