Job Market paper

The power of good neighbors

An analysis of intergenerational mobility


This paper provides a theoretical and quantitative analysis of the role of good neighbors (i.e., neighbors with better information signals about the current state of the economy) for saving and human capital formation outcomes. It proposes a framework that links the spillover effects of optimal neighborhood selection with inequality and network analysis and estimation.  This research contributes to the growing literature on intergenerational mobility. Theoretically, it introduces a novel network-based model of information cluster formation, incorporating endogenous neighborhood selection. Living in neighborhoods with high connectivity produces information spillovers, so that better financial choices result from those connections and are ultimately reflected in savings decisions and wealth accumulation. Similarly, the spillover affects the choice of parents on their children’s education, further enhancing wealth accumulation over time. A network is more homogeneous (heterogeneous) when the agents are more (less) connected at the same information level. The study characterizes the unobservable parameter values, returns of investment and precision signal of the agents, for which there is a stable homogeneous outcome that does not maximize the sum of expected returns of all the agents. The unobserved parameters are estimated with data from the Opportunity Atlas. Quantitatively, the compensating value in welfare for a low-information household (an agent with a less precise signal) of living in a neighborhood with 25% of high-information neighbors, instead of none, is estimated to be roughly 16% of the median saving. Furthermore, the simulation based on the estimated parameters in the network structure predicts a ten-percent decrease in Economic Connectedness – the fraction of high-informed friends among low-informed individuals – for the next generation, compared to the benchmark of a random network.



Link to folder with Sildes and WP: