Hiring and Learning in Online Global Labor Markets

Abstract

This paper uses data from freelancer.com – an online platform that\nallows employers and freelancers to search for, and match with, each\nother – to study the effect of freelancers’ country of\norigin on their likelihood to be hired. Having to rely on a relatively\nsmall number of characteristics, employers use the freelancer’s\ncountry of origin to infer the expected service’s quality. This\nsetting also allows me to document how employers’ experience in\npast hires affects their behavior in current hires. I find that\nfreelancers from developing countries are less likely to be hired when\nthey have no individual reputation, and as individual reputation becomes\nbetter this country effect disappears. I show that following a good\nmatch with a freelancer, employers are more likely to hire freelancers\nfrom the good match’s country. These these findings are consistent\nwith statistical – rather than purely taste-based – discrimination.