[krageon]

The literature here is of poor quality. For a salient example, look at this: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3020503

Where a giant group is compared to an extremely tiny group and the differences between them are used to make sweeping generalizations about everyone. When this is the quality to expect, it's just good practice to be skeptical.

Edit: Even if this weren't the case, the enormous spread of estimates when it comes to revenue impact really says it all.