A0651
Title: Nonparametric discrete-choice and sensitivity to distributional assumptions on random utility
Authors: Jonas Lieber - Imperial College London (United Kingdom) [presenting]
Alexander Torgovitsky - University of Chicago (United States)
Pietro Tebaldi - Columbia University (United States)
Abstract: New results are introduced to estimate sharp nonparametric bounds on pre-specified demand counterfactuals in a discrete choice model, avoiding assumptions on the distribution of random utility. The set of inequalities characterizing choice under quasilinear utility points to an efficient linear algebra procedure to partition the space of unobservable, maintaining all relevant information for the question of interest. The method is used to discuss the sensitivity of parametric estimates to the relaxation of widespread distributional assumptions on random utility.
