This is in response to the stumbling block for the OAI: that institutions can't set up open access archives containing the work of their researchers because they don't own the copyright on the postprints, which are what readers are interested in.
Instead of institutions paying open-access journals the cost of peer review and distribution, they could pay commercial journals a one-off licence fee that would allow them to distribute the institution's publications in that journal from a local archive.
Update: Self Archiving FAQ ("Mandate that the full digital text of all refereed publications should be deposited in the University EPrint Archives") → ECS Research Self-Archiving Policy ("Many publishers will allow the peer-reviewed postprint to be archived") → Publisher copyright policies & self-archiving (Publishers who allow self-archiving in institutional archives - Elsevier: NO; Nature: NO; Springer: NO; John Wiley: NO; Blackwell: NO).
See also: Open Access Reprints