Discrimination, Sickness Absence, and Labor Market Policy

Abstract: This dissertation consists of an introduction and four self-contained essays:Essay 1 (with Stefan Eriksson) investigates empirically whether being unemployed per se reduces the probability of getting contacted by a firm. Individuals registered at the Swedish employment offices post their qualifications in a database available to employers over the Internet. Since we have access to exactly the same information as the firms, we can minimize the problems associated with unobserved heterogeneity. Our results show that an unemployed applicant faces a lower contact probability, and receives fewer contacts, than an otherwise identical employed applicant, thus supporting the notion that firms view employment status as a signal for productivity.Essay 2 evaluates an experiment where employees at randomly chosen establishments received half a day off if they completed a full calendar month without any sick-leave. Using individual panel data, the absence rates of these individuals are compared to the absence rates of individuals at establishments with no such program before, during and after the treatment periods. Overall, the bonus caused a sharp reduction in absenteeism, especially for women, highly educated individuals and part-time workers. Essay 3 (with Per-Anders Edin) provides evidence on discrimination in the hiring process. We use Internet data generated from a “policy experiment”, in which individuals can choose not to reveal their name and gender to potential employers. By comparing the “contact rate” of censored and non-censored women and minorities, we find that women have a 20 percent lower chance than men of getting contacted by employers and that this differential is fully explained by discrimination. Our results concerning ethnic discrimination are less conclusive, probably due to measurement errors.Essay 4 examines if and how the personnel at the Swedish Employment Office matter. The analysis shows that caseworkers explain a substantial part of future outcomes in terms of employment status and earnings, but have no significant effect on wages. Caseworkers that send their clients to classroom training or on-the-job training are less successful than caseworkers that provide basic job-search assistance. If caseworkers’ preferences in previous years towards treatments are uncorrelated with unobserved present working strategies, these estimates correspond to causal treatment effects.