DP11675 Preference Estimation with Unobserved Choice Set Heterogeneity using Sufficient Sets

Author(s): Gregory S. Crawford, Rachel Griffith, Alessandro Iaria
Publication Date: December 2016
Date Revised: October 2017
Keyword(s): attention, discrete choice demand estimation, endogenous product choice, search, sufficient sets, unobserved choice set heterogeneity
JEL(s):
Programme Areas: Industrial Organization
Link to this Page: cepr.org/active/publications/discussion_papers/dp.php?dpno=11675

This paper introduces "sufficient sets'' and shows how they can address unobserved choice set heterogeneity. Motivated by economic theory, a sufficient set is a combination of a consumer's observed choices that is a subset of their true but unobserved choice set. We introduce a Sufficient Set logit that "differences out'' consumers' true choice sets, permitting consistent parameter estimation, and show how to use sufficient sets to semi-parametrically estimate preferences and lessen the computational burden of methods that integrate over the distribution of choice sets. We illustrate our ideas using Monte Carlo simulations and by estimating demand for chocolate in the UK.