Employee referral programs (ERPs) are randomly introduced in a grocery chain. Larger referral bonuses increase referrals and decrease referral quality, though the in- crease in referrals from having an ERP is modest. However, the overall effect of having an ERP is quite sizable, reducing attrition by 15% and decreasing firm labor costs by up to almost 3%. This occurs, partly, because referrals stay longer than non-referrals, but, mainly, because non-referrals stay longer in treated stores than in control stores. The most-supported mechanism for these indirect effects is that workers value being involved in hiring. As a result of the RCT, the firm rolled out an ERP to entire firm. For non-grocery jobs, which are regarded as more attractive, direct effects of ERPs are larger than during the ERP, but indirect effects remain substantial. |