Here's Elsevier's latest revision of the wording of its author rights agreement stating what rights Elsevier authors retain for their "Accepted Author Manuscript [AAM]".

Elsevier believes that individual authors should be able to distribute their AAMs [Accepted Author Manuscripts] for their personal voluntary needs and interests, e.g. posting to their websites or their institution’s repository, e-mailing to colleagues. However, our policies differ regarding the systematic aggregation or distribution of AAMs... Therefore, deposit in, or posting to, subject-oriented or centralized repositories (such as PubMed Central), or institutional repositories with systematic posting mandates is permitted only under specific agreements between Elsevier and the repository, agency or institution, and only consistent with the publisher’s policies concerning such repositories. Voluntary posting of AAMs in the arXiv subject repository is permitted.

Please see my prior analyses of this Elsevier double-talk about authors retaining the right to make their AAMs OA in their institutional repositories "voluntarily," but not if their institutions mandate it "systematically." Here's a summary: 

1. The author-side distinction between an author's self-archiving voluntarily and mandatorily is pseudo-legal nonsense: Authors can truthfully safely assert that whatever they do, they do "voluntarily." 

2. The institution-side distinction between voluntary and "systematic" self-archiving by authors has nothing to do with rights agreements between the author and Elsevier: It is an attempt by Elsevier to create a contingency between (a) its "Big Deal" journal pricing negotiations with an institution and (b) that institution's self-archiving policies. Institutions should of course decline to discuss their self-archiving policies in any way in their pricing negotiations with any publisher.

3. "Systematicity" (if it means anything at all) means systematically collecting, reconstructing and republishing the contents of a journal -- presumably on the part of a rival, free-riding publisher, hurting the original publisher's revenues; this would constitute a copyright violation on the part of the rival systematic, free-riding publisher, not the author: An institution does nothing of the sort (any more than an individual self-archiving author does). The institutional repository contains only the institution's own tiny random fragment of any individual journal's annual contents.

All of the above is in any case completely mooted if an institution adopts the ID/OA mandate, because that mandate only requires that the deposit be made immediately, not that it be made OA immediately. (If the author wishes to comply with a publisher OA embargo policy --which Elsevier does not have -- the repository's "Almost-OA" eprint-request Button can tide over researcher needs during any OA embargo with one click from the requestor and one click from the author.)