Identification and Estimation of Discrete Choice Models with Unobserved Choice Sets

We propose a framework for nonparametric identification and estimation of discrete choice models with unobserved sets. recover the joint distribution sets preferences from panel dataset on choices. assume that either latent are sparse or is sufficiently long. Sparsity requires number possible to be relatively small. It satisfied, instance, when nested, they form partition. Our procedure computationally fast uses mixed-integer optimization support Analyzing ready-to-eat cereal industry using household scanner dataset, we find ignoring unobservability can lead biased estimates due significant heterogeneity in