Systems without bias still show differential outcomes

This rerun was playing on NPR the other day. From What Can Uber Teach Us About the Gender Pay Gap? (Ep. 317) by Stephen J. Dubner. Draws heavily on the research of Cody Cook et al. which I have earlier commented on but pulls in other researchers as well.

Basically, in bias free environments and working the same number of hours, women earn 7% less than men due to 1) differences in the time and location (20% of the gap), 2) differences in how productive they are (number of tasks completed per hour) (30% of the gap), and differences in fast they drive (50% of the gap). A 2% difference in driving speed translates into a 3.5% difference in income.