Public Economics and Policy Seminar - Michael Siegenthaler (ETH Zürich)

Women (compared to men) and ethnic minorities (compared to natives) face inferior labor market outcomes in many economies, but the extent to, and the channels through, which these are caused by discrimination remains unclear. We deploy a new approach to quantify hiring discrimination by tracking recruiters’ search behavior on employment websites and using supervised machine learning to control for all job seeker characteristics visible to recruiters. Applying this methodology to the Swiss government-affiliated recruitment platform, we find that depending on origin country, ethnic minorities face between 3-19% lower contact rates compared to otherwise identical natives. For gender, effects are highly heterogeneous too: women face a penalty of up to 40% in male-dominated professions, and the opposite pattern emerges for men in female-dominated professions. Our approach provides researchers and policy-makers with a widely applicable, non-intrusive, and cost-efficient tool to continuously monitor hiring discrimination and to suggest policies to counter it.