Title: "Identifying Peer Effect Structure in Models of Discrete Choice "
|Date||Friday 9 February 2018|
|Venue||Heriot-Watt, Room MB1.62|
Speaker: Rob Hicks (College of William and Mary) – joint work with Arnab Bhattacharjee and Kurt Schnier (University of California Merced)
In this paper we outline a method for estimating peer effects amongst agents in the context of a discrete choice model and examine the case commonly encountered when studying firm behavior when peer networks may exist but membership is not observed. We apply commonly used models of spatial discrete choice and extend them to capture these unobservable peer effects and discuss the conditions under which peer networks are identified in a Random Utility Model setting. The Bayesian estimating framework we employ allows for the estimation of peer effects alongside the typical random utility site choice model. We perform a Monte Carlo analysis to investigate model performance under various network structures.