UCLA

This dissertation consists of three main chapters that study social interactions in networks. In Chapter 1, I study a market with many-to-many contracts when the number of market participants increases. Many-to-many contracts allow a seller to trade with multiple buyers and a buyer to trade with multiple sellers. I focus on investigating the identification of payoff parameters through data observed from equilibrium matches in a large many-to-many matching market. In many-to-many matching markets, several issues have to be addressed: bias would arise since the outcomes are only observed when links are formed between two agents, and the maximum number of relationships an agent can enter into would possibly affect the set of stable outcomes. I show that under certain conditions, the number of firms (workers) that are willing to be matched to a specific worker (firm) grows at a rate regardless of the capacity of both sides. Furthermore, I show a correspondence between the stable matching outcomes in a many-to-many matching framework and that in a one-to-one matching framework.

In Chapter 2, I conduct a structural econometric analysis of the diffusion process with players who observe their neighbors and make decisions based on their neighbors' decisions. I study the identification and estimation of diffusion processes in social and economic networks. Compared to the classic econometric diffusion literature that assumes a continuous population with a stochastic network structure, I provide a new econometric framework to analyze diffusion processes in fixed networks where Bayesian players observe their close neighbors. I demonstrate the existence of the equilibrium of the model and characterize the unique symmetric equilibrium. Based on these theoretical findings, I propose a consistent and tractable two-step estimator for payoff parameters using feasible data from a single large network. I evaluate the finite sample performance using Monte Carlo simulations.

Chapter 3 applies the network diffusion model to data on the participation of a microfinance program in Indian villages to describe the impact of neighbors on individual decisions. Our model allows us to study the various network effect across different types of agents who care about their neighbors' opinions. It depends on unknown equilibrium beliefs, which specify agents' expectations about their neighbors' decisions. Using participation data from 43 villages, each including about 200 villagers, I estimate these equilibrium beliefs and the network effects.