Path to equality: Economic implications of referral hiring and social networks

I investigate how the flow of job information in social networks affects labor market outcomes. The initial chapters of the thesis concentrate on how propensities to hire workers through social channels, and the structure of social networks affect levels of individual and group inequality, and firm level segregation. Regarding inequality, if population groups have a homogeneous social structure, referral hiring does not create steady state inequality at the level of groups. However, referral hiring does slow the rate of convergence to a steady state. Inequality can be created if social networks have heterogeneous properties across individuals or groups. Individuals do better with more random social networks. Individuals benefit from access to wide bodies of information provided by reaching more labor market contacts through random networks. But groups prefer less random social networks. Groups benefit from non-random networks by being able to restrict access to job information to group members. Thus the definition of good networks in a labor market context depends on whether one has an individual or a group level perspective. The difference in incentives between individuals and groups creates a collective action problem. Individual members of a defined group will create more randomness in networks than the group would prefer.;With regard to workplace segregation, referral hiring creates group level segregation of workers across firms if social networks are segregated. As referral hiring decreases or as social networks become more random, workplace segregation decreases. I use data on the staff of colleges and universities in the United States to measure the amount of workplace segregation that can be explained by referral hiring. The results suggest that referral hiring accounts for approximately one-half of the observed levels of gender segregation in the data.;The final chapter of the thesis considers the endogenous creation of referral networks in a labor market model. A local selection evolutionary algorithm is employed to study the adapted graph topologies that result. The evolved networks display mixtures of regularity and randomness, consistent with small-world networks.