Essays in network formation

This thesis contains three essays on the formation of social networks. The first essay focuses on citation networks while the second one proposes a general model of network formation. The third essay is an empirical investigation of a person's kinship network.;Chapter 2: Competing to be a 'Star': A model of sequential network formation. This paper studies a sequential-move game of network formation with capacity constrained agents. The sequential nature of the game allows us to focus on the role played by foresight in the formation of particular network configurations. We find that the equilibrium network structure depends crucially on the rate of decreasing returns (decay) in the payoffs. With homogeneous agents and a capacity of one link, the equilibrium network is a fully connected graph for extreme levels of decay. The architectures for very low and very high levels of decay are a complete star and a graph similar to the hub-and-spoke architecture, respectively. For intermediate levels of decay, however, the network configuration might have more than one component. This occurs because agents have the incentive to reject a link in some subgame in order to become central in the network at some future date. With heterogeneous agents, the equilibrium network for both very high and very low levels of decay is a star with a high ability agent as the center. But it is possible to have structures where the high ability agent is kept isolated by the rest of the society comprising of low types. This depends on the decay factor and the difference in the abilities of agents. The equilibrium is, for most values of decay, inefficient. Hence, we can possibly explain the existence of multiple small groups in various social settings even when a single cohesive network would be more efficient.;Chapter 3: Family Embeddedness: An Empirical Investigation. Using the Netherlands Kinship Panel Data, we investigate the relationship between an individual's embeddedness in family ties, effort exerted on the job and social attitude. In particular, we want to explore whether stronger familial ties are associated with any negative impact on personal incentives or on society. We find that agents with a higher level of family ties at birth maintain a higher level of ties at a later stage. They also have a higher involvement in volunteer work and are more tolerant of migrants. Moreover, the level of inherited ties affect a person's current work effort positively. This is possible due to peer effects or due to a correlation with the occupational choice of individuals. Other factors which influence both ties and effort have an opposite effect on the two. Hence, the level of initial embeddedness in family, by itself, does not have negative effects on social attitude or effort. However, there possibly exists a negative relationship between current ties and effort due to family obligations or time constraints. We also find that negative social attitudes could be nurtured by higher income or wealth.;Chapter 1: Citations and the Diffusion of Knowledge: An Economic Analysis (with Kalyan Chatterjee). Citation patterns in many academic disciplines have displayed a pattern of connections similar to those observed in many other different real-world contexts, such as links on the world-wide web. The various models that have been proposed to generate such networks, generically called "preferential attachment models", rely solely on random link formation and copying and do not take into account rational choice among authors in an academic community, which would consider the competition for citations and ensuing professional success. In this paper we construct such a model with rational agents to understand some aspects of citation patterns and knowledge diffusion in a specific academic field. We show that rivalry or competition in citations might be an obstacle to diffusion, depending on behavioral rules specific to the field. Increased heterogeneity in the quality of papers reduces this effect. After considering models with complete information, we analyses models with private information about quality of one's own paper and use the framework to consider the interaction of this process with acquaintance networks and strategic entry. Superimposing the citation process on an acquaintance network yields patterns different from preferential attachment. Strategic entry leads to cascades of papers. Though we might have ex-ante efficiency in some equilibria, ex-post efficiency is not guaranteed. Ex post efficiency cannot be guaranteed since it is always possible in equilibrium that a good paper "dies" and a worse one survives, but ex ante efficiency is sometimes attainable.