Essays in the Economics and Econometrics of Networks and Peer Effect

• Department of Economics

PhD thesis defence by Zheng Wang

This thesis contributes to the understanding of peer effect, both methodologically and empirically.

In the first chapter I propose a new causal framework to study the effect of peer relationships. Moreover, an identification strategy based on unconfoundedness is provided for peer networks that are endogenously formed, such as friendships and production networks. Thanks to the nature of network data, confounders can be inferred from the adjacency matrix, and therefore the identification does not require the assumption that all confounders have been observed. The identification strategy suggests the use of propensity score based estimators, which means estimation can be easily and flexibly done with existing statistical packages. To show how this works in practice, I re-analyse two empirical papers on peer effect by Cools et al. (2019) and Olivetti et al. (2020). Instead of studying the effect of peers within one's cohort on student's long term outcomes, I study the effect of friendship. I find that close social interactions with friends have very different impact on one's life than general interactions from cohort peers.

The second chapter lays out extensions to the framework proposed in Chapter 1. These extensions allow researchers to use the framework more flexibly in empirical applications. For readers more interested in the technical aspect of the previous chapter, it also provides proofs to lemmas and propositions proposed in the previous chapter. Finally, this chapter includes more detailed results of the simulation exercise and the empirical application.

The final chapter studies the effect of competition difficulty on one's effort exertion. Using data from Duolingo leaderboard where language learners are randomly put into group competition, I find evidence suggesting that people react to competition difficulty differently depending on their time constraint and their level of commitment.

The event will be HYBRID and will also be streamed via Zoom.