How does collective reputation affect hiring and selection into jobs? Using detailed hiring data from a global online labour market, where the country of residence is the salient group characteristic, we document a mechanism through which collective reputation can perpetuate initial group inequality. Using an instrumental variable strategy, we first identify reputational externalities between an employer's very first hire and the propensity to contract more workers from the same country. Employers, contingent on their first worker's performance (as measured by a public rating), go on to almost exclusively hire from the same country. This coincides with a strong and positive supply-side sorting response: Observing their predecessor's success, workers from the same country are more likely to apply and tend to be of higher quality. Employers, facing better applicants, are in turn more likely to continue providing top ratings for later hires from the first hire country. Collective reputation hence appears to serve as a coordination device that enables workers to positively sort with employers: Good workers then attract more good workers from the same country and vice versa. JEL codes: J7, J16, L14 * I am indebted to Oriana Bandiera and Robin Burgess for their supervision and continued guidance. I