Do Job Networks Disadvantage Women? Evidence from a Recruitment Experiment in Malawi

Using a field experiment in Malawi where applicants took part in a competitive selection process to become qualified for future surveyor positions with a local firm, we find that highly skilled women are systematically disadvantaged through the use of referrals. This happens both because men choose to recommend other men, and because women on average refer fewer candidates who qualify for the position. Our experimental design allows us to shed light on the network characteristics which lead to this disadvantage. Both men and women are connected to men and women at equal rates, so that the biases in referral behavior are not caused by segregated networks. By developing a model of how individuals choose referrals from among their social network and exploiting random variation in the terms of the referral contract, we document that both mens’ and womens’ biases result from social incentives rather than expectations of performance. Moreover, we also document that screening is maximized when men refer men. Therefore, employers who use networks to grant non-pecuniary benefits to employees or to screen new hires would have an incentive to encourage this disadvantage. This paper suggests that the use of social networks in hiring creates an additional channel through which women are disadvantaged in the labor market.