In this note we study the conditions under which leading models for underreported counts are identified. In particular, we highlight a peculiar identification problem that afflicts two of the most popular models in this class.