How Do Workers Choose Referrals?

Globally, a large fraction of workers are hired through job references. An extensive literature, both in sociology and in economics, has documented the impor­ tance of social connections as a means of job search. For example, 30-60% of jobs in the United States are found through referrals. In Kolkata, India we find that 45% of our sample has found a job for a friend. A large and grow­ ing empirical literature in economics has more recently utilized a variety of quasi-experimental variations in the size or quality of job networks to con­ firm what survey data has shown: Job networks are empirically important. While it is clear that job networks matter, we know much less about how they work and what the consequences of the use of informal social connec­ tions in the hiring process are to either firms or unemployed job seekers. In fact, while we have substantial evidence that job referrals are important in labor markets, we have much less evidence on even very basic questions about how they work, such as whether the use of referral systems results in good hires. Theory has suggested two rea­ sons to suspect that networked hires may be high ability: first, high-ability workers may associate primarily with other high-ability people, so that their referrals are mechanically high quality. Second, if networks are het­ erogeneous, then employees may be willing and able to screen on their employer’s behalf. That is, if they know which of their friends represent a particularly good fit for a job open­ ing, they may be willing to use that information and make a referral. In a recent paper forthcoming in the American Economic Review, we test whether employees have useful infor­ mation about their network members, and under what conditions they are willing to use that information. In our view, both the capacity and willing­ ness of employees to screen on their employer’s behalf can be questioned. Social networks are formed for many reasons, at least some of which are unrelated to a person’s poten­ tial as an employee for a specific job. Moreover, in order for employees to be useful in choosing which of their friends would make the best employee, it must be the case that whatever incen­ tive they have to screen on the part of the employer can overcome what­ ever incentives they have to use that job opportunity in a different way. For example, in much of the lit­ erature on networks in developing countries, the role of networks to insure against risky events would sug­ gest that people may make claims on the next good opportunity—like a job opening—that their friends learn of, and offer in exchange for a promise of sharing the next good opportunity that they learn of. This possibility would mean that new job referrals go to people who have the strongest claim to the next good opportunity rather than to the person who would be best at the job. Moreover, it is surely the case that altruism plays an important role in social networks, giving incen­ tives to refer the network member one cares about the most rather than the friend who you believe to be the best worker or best match for that job.