In most empirical studies of discrete choice, the presence of an outside good is assumed.
Can you give an example of a (well published) empirical paper, where there is no outside good and thus utility normalization is w.r.t. to any of the inside goods?
