Experimental Design
We work with a recruiting agency to randomize which candidates are included on lists of work-seekers sent to firms, generating random variation in the probability of hiring. We estimate treatment effects on candidates’ employment, earnings, skill development, and job ladders. We also use random variation in the types of workers hired to build a predictive model of worker productivity, incorporating detailed baseline skill assessments, and assessing which hiring rules maximize average worker performance and gender inclusivity.