| This thesis explores aggregate implications of social interactions, in particular the relationship between social interactions and multiple equilibria, thresholds, and tipping behavior at the neighborhood level. In this dissertation, I apply this framework to analyze some issues in unemployment and teen smoking.; In "A dynamic model of job networks and persistent inequality," I characterize the effect of informal job networks on the distribution of employment across neighborhoods. My model formalizes the well-documented phenomenon that employed friends often provide referrals to potential employers, increasing one's employment probability. I find that this interdependence in employment generates multiple equilibria and the associated phenomena of tipping and thresholds, as well as other interesting nonlinear dynamics.; In "Social interactions, thresholds, and unemployment in neighborhoods," I demonstrate the empirical relevance of unemployment thresholds in neighborhoods. Using data on Census tracts in twenty United States cities, I find that when the percentage of college graduates within a neighborhood falls below about twenty, the predicted unemployment rate in that neighborhood rises dramatically. To investigate whether this pattern is the result of neighborhood sorting rather than true social interactions, I develop a simple model in which individuals sort into neighborhoods based on long-run productivity. I then estimate the implied distribution of long-run productivity if this sorting is to generate the observed nonlinearity in the reduced form relationship between human capital and unemployment, and show the implied distribution is inconsistent with the likely distribution of long-run productivity in the population. This suggests that sorting alone is not enough to explain the observed nonlinearity, providing evidence for social interactions.; The third chapter, "Do social interaction effects explain the race differential in youth smoking?" addresses a public health puzzle---during the late 1980's smoking rates among black teenagers fell dramatically, while smoking rates among white teenagers remained stable. This pattern can be interpreted as movement between multiple equilibria, and the paper evaluates this interpretation empirically. While peer behavior is closely related to youth smoking, I find no evidence for multiple equilibria. |
