Discrimination in hiring : Some experiments, perspectives, and implications

Abstract: Hiring discrimination is illegal, morally distasteful, and seen as incommensurate with modern societal ideals. From an economic perspective, if employers hire based on anything other than an applicant’s expected productivity they are behaving inefficiently. If group markers are imperfect signals of productivity, discrimination is also inefficient. Measuring discrimination is a substantial practical challenge but indispensable to policy development and theory evaluation. This thesis focuses on correspondence testing, experiments where researchers create fictitious applicants who apply for real jobs, and then analyze differences in responses between groups to arrive at credible estimates of discrimination. In chapter I, "Do ethnicity and sex of employers affect applicants' job interest? An experimental exploration," co-authored with Ali Ahmed and Niklas Ottosson and published 2020 in Journal for Labour Market Research, we present the findings of a survey experiment. We tested the novel hypothesis that job seekers may discriminate against employers based on ethnicity or gender when they are choosing jobs to apply to. Ultimately, we concluded that the survey experiment provided no evidence of such discrimination. In chapter II, "Hiring discrimination against transgender people: Evidence from a field experiment," co-authored with Per A. Andersson and Ali Ahmed and published 2020 in Labour Economics, we present the findings of a correspondence experiment that tested for hiring discrimination against transgender applicants. We found that transgender applicants were indeed discriminated against in hiring, but that there were some important nuances. For example, transgender men seemed to be discriminated against in male-dominated occupations because they were transgender and in female-dominated occupations because they were men. In chapter III, "Gender discrimination in hiring: An experimental reexamination of the Swedish case," co-authored with Ali Ahmed and Shantanu Khanna published 2021 in Plos One, we present the findings of a study that combined data from three previously published correspondence experiments. Although these experiments were originally designed to test other hypotheses, we used the data to test for gender discrimination in hiring. We found discrimination against males, largely driven by female-dominated occupations. In chapter IV, "An assessment of the correspondence testing methodology," I describe and analyze the methodology and ethics of correspondence tests. I do this by reviewing the 199 correspondence studies published between 2005 and 2020, focusing on methodological choices and the ethical implications of those choices.