As a preliminary observation, I will just note that this phenomenon is a good reason that academic journals should practice blinding of papers sent to peer reviewers. For journals that practice effective blinding of papers, this whole situation cannot arise, so we are really just talking about the journals that don’t use blinding, or don’t do it effectively.
I think some of the answers here are being a bit uncharitable to your question, by making the assumption that deception of the peer reviewers in the process must be coupled with some broader deception of the journal, or even ultimate publication under a false name or affiliation. Ethical problems arise in those latter actions, but they are not necessary for what you want to achieve here. One can imagine an approach in which an author submits a paper to a journal (that does not use blinding) using their correct name and affiliation in the submission system and with correspondence with the editor, but giving a false name and affiliation on the actual document that will be seen by the peer-reviewer. In this case the submitter would presumably be open with the editor about what they’re doing and why they’re doing it. If this were drawn to the attention of the editor, it would then be up to the editor to make a decision on whether a deception of their peer reviewer is justified or not. (In practice I doubt a journal editor would agree to this, since they can achieve the same results by blinding papers, without any use of deception.) If the editor is willing to go along with it, one can even imagine further safety protocols being added, such as disclosing the true author of the paper to the reviewer after the review is complete, and giving them some debriefing on the deception and why it occurred. And obviously if the paper were to be accepted and published, it would be published with the correct author name and affiliation, not the one used for deception in the peer review.
If such an approach were to be proposed, it would be quite similar to a number of experiments in academia where a subject group is exposed to a temporary deception to affect their behaviour and then they are debriefed on that deception at the end of the experiment. Ethics panels at universities routinely examine proposals like this to determine whether or not they meet ethics requirements. Typically, such actions will be allowed so long as the deception has a bona fide research purpose, is not too harmful, and so long as the people being deceived are debriefed at the end to be made aware of the deception and its purpose. In the case of a submission of a paper to a journal the difference here is that this is not an experiment, and the only research purpose is to shield existing research work from an adverse peer review effect that is not being dealt with adequately by blinding.
While such a proposal might not be looked on kindly, I’m not convinced that it would be impossible to do this ethically, and certainly without distortion of publication statistics. Nevertheless, the whole idea is somewhat undercut by the fact that it is simpler to just give a blinded paper to the peer reviewer to achieve essentially the same goal.