I study a reform that shifts public lunchtime childcare in kindergartens and schools from an opt-in to an opt-out default. With the new default, children are automatically enrolled in low-cost lunchtime childcare unless parents opt out. Leveraging the staggered rollout in Zurich, Switzerland, in an event-study design, I document that the reform increases public childcare enrollment by 39%. The reform increases the employment rate of mothers with children in kindergarten by 4.8% and their earnings by 7.5%. I provide suggestive evidence that the reform reduces norm-related behavioral frictions by making public lunchtime childcare the initial arrangement for all families.

I study role model interventions that expose high school students to talks by STEM professionals during single-day school events. Combining data on more than 1,500 speakers with administrative records on 240,000 Swiss high school graduates, I find that exposure in the final year of high school increases STEM college enrollment by 4.1% and STEM degree completion by 7.7%. Events with a higher share of female speakers have larger effects for both female and male students. Presentation style explains part of these effects: female speakers are more likely to use interactive formats, to which students respond positively. Accounting for interactivity reduces the female-speaker effect by 37%.

I investigate how mentoring programs shape female students' educational choices. Collaborating with the Swiss Academy of Engineering Sciences, this ongoing work investigates the impact of a seven-month mentoring program that pairs female students with female STEM professionals. Exploiting oversubscription, I randomize access to the program and assess its impact on students’ college educational choices, tracked through administrative data. Two cohorts of a total of 460 students completed the baseline survey and were randomized into the program in June 2024 and 2026, respectively.

We leverage large-scale clickstream data from a major online recruitment platform to systematically estimate the returns to human capital for minority and majority jobseekers. We track the search behavior of recruiters on the recruitment platform and exploit that we observe the whole selection process and the same information about the jobseekers as recruiters do.