Econometrics of Network Analysis

Catalog description

An overview of the models and methods for analyzing data with cross-sectional dependence, i.e., those able to explicitly test behavioral models with interdependent agents' decisions. The technicalities are presented in a basic formulation, favoring the transmission of ideas, intuitions, and stressing the links with underlying behavioral mechanisms essential to guiding the interpretation of the results. The open questions in the economics literature are emphasized. They include: 1) the definition of the reference group; 2) the possible presence of unobserved attributes that may generate a problem of confounding variables (spurious spatial correlation); and 3) simultaneity in agents' behavior that may hinder identification of exogenous effects, i.e., influence of agents' attributes) from endogenous effects, i.e., influence of agents' outcomes. This short course focuses on identification issues.