Quantifying Spillovers on an Online Social Network

Speaker: Dr. Sean Chu
Affiliation: Facebook

Abstract: We empirically model the formation and dissolution of friend ties on Facebook as a non-cooperative game of network formation. In this game, agents make sequential decisions with respect to individual friend ties, and the potential value of mutual friend ties introduces network externalities. Crucially, our model includes hidden state variables that allow for unobserved heterogeneity in the value of individual friend ties, and that allow us to accommodate the fact that not all of an agent’s decisions necessarily result in an observed change to the friend graph. We identify our model by exploiting the panel structure of our data, and estimate our model using a computationally tractable two-step estimator combined with an EM algorithm.

Biography: Sean Chu is an industrial organization economist on the Data Science team at Facebook. His research focuses on empirically modeling games of incomplete information, dynamic games, and other aspects of individual and household behavior. At Facebook, he is applying these tools to problems relating to network behavior, peer effects, advertising efficacy, and online experimentation. Before coming to Facebook in 2014, Sean worked for seven years at the Federal Reserve Board of Governors, where he researched household behavior in credit markets and the role of information asymmetries and imperfect competition in financial markets. Sean received a Ph.D. in Economics from Stanford, and wrote a dissertation on product bundling and non-price competition in consumer markets.

For more information, contact Prof. Mihaela van der Schaar ()

Date/Time:
Date(s) - Feb 29, 2016
1:00 pm - 2:00 pm

Location:
E-IV Tesla Room #53-125
420 Westwood Plaza - 5th Flr., Los Angeles CA 90095