Having formally stated since
2004 that Elsevier
authors retain the right to self-archive their final, refereed, revised, accepted drafts, unembargoed, in their institutional OA repositories, Elsevier has
tried to imply that the author's
institution somehow does not have the right to
host that draft, if it mandates self-archiving!.
And Elsevier tries to bring this up, not in its public negotiations with its authors, but in its confidential private negotiations with institutions, in the context of Big-Deal pricing agreements.
(Institutions should of course politely decline to discuss university self-archiving policy in any way, in their journal pricing negotiations with Elsevier or any other publisher.)
Just as it makes no sense (hence carries no legal force) to say to Elsevier authors that they retain the right to self-archive in their institutional repositories "voluntarily" but not "mandatorily," it makes no sense to say that if authors' institutions mandate self-archiving, then they may not host the self-archiving
that their Elsevier authors formally retain the right to do in their institutional repositories.
And "systematicity" as a grounds for over-riding the Elsevier author's retained right to self-archive unembargoed is just as nonsensical (hence non-binding) at the host-institution level, particularly since Elsevier has (wisely) conceded the right of all its authors to self-archive in arXiv (as they have been doing since 1991).
Arxiv, a global repository in which close to 100% of the articles in several subdisciplines of physics, mathematics and astrophysics are accessible
is indeed a systematic collection of Elsevier journal content.
But no individual institution, hosting its own tiny, arbitrary fragment of global journal output can be faintly construed as a "systematic collection" of Elsevier content, any more than any individual author's collection of his own self-archived articles can be.
So here too, Elsevier authors can and should safely ignore the FUD and double-talk and self-archive their final drafts in their institutional repositories immediately upon publication, just as they have been doing since 2004.