How to Compute the Largest Number of Rationalizable Choices

Abstract

When checking revealed preference conditions on experimental or field data, a commonly reported measure of choice consistency is a function of the largest number of rationalizable choices. While there are known methods to compute the largest number of rationalizable choices for the standard consumer problem with linear budgets, computation of this object is unknown for general budget sets. In this paper, I provide a mixed-integer linear program that facilitates computation of the largest number of rationalizable choices for general budget sets.