This is the usual story: Once a free or reasonably priced journal is successful, it is bought, prices are raised, and access restricted. The lure of money is too tempting.
The most recent case concerns the BEPress journals that had pioneered open access in economics (well, actually quasi-open access, but this was acceptable). Aaron Edlin has sold them to DeGruyter, and quasi open access turned into gated access. The author’s rights are disregarded, of course.
We see here that well-defined property rights might bring about economic inefficiency: If a journal can be sold, it will be sold and turned into a goldmine, even if this is inefficient from an economic point of view. This can never happen to RePEc, as it is not owned by anybody. Under the presumption that open access is economically more efficient than gated access, ill defined property rights contribute to efficiency.
As an author make sure that you publish in a journal that cannot be sold, or is unlikely to be sold. Perhaps the existing free journal software should carry a clause that free use is permitted only for open access journals, and other uses are permitted only on paying a stiff fee. This would make credible to the authors that their work remains accessible and would solve the problem even with well-defined property rights, but this is unlikely to happen.
The RePEc Blog 
http://archiv.twoday.net/stories/326205765/ Wrong recommandation. The right would be: make sure that your papers are CC-BY. Thus they can be mirrored without problems.
This is usually not possible, as the publishers insist on their contract terms. Given this fact, it is advisable to publish a pre-print on the Net at a service like MPRA (http://mpra.ub.uni-muenchen.de/). This leaves you the copyright to the pre-print and keeps your work openly accessible, although not in its final form.