I model the emergence of job referral networks as an endogenous response to local labor market conditions. The density of the job referral network workers use to share job information is determined in general equilibrium, along with unemployment, vacancies, and earnings. To make the referral network endogenous, I introduce the concept of network balance as an equilibrium condition linking individual search decisions to aggregate network structure. In equilibrium, referral networks are most

more »

... at moderate levels of labor market tightness -that is, in labor markets where job information is hard, but not impossible, to come by through formal search. Using data on individual referral use from the Cornell National Social Survey, I provide evidence in support of the model's counterintuitive assumption that workers are less likely to find jobs through referral in markets where referrals are more widely used. JEL Codes: J64, R23.