While I was at UCLA, I heard some fairly strong pushback against institutional repositories. Much of it didn't trouble me at all; it's no more than what I've said myself. One strain, however, did bother me rather.
It went something like this (and I apologize; I am probably traducing it): How do libraries justify spending on open access—making local materials available to the world—if our guiding mission is to buy appropriate materials specifically on behalf of our patron base?
To make that abstract question more concrete, consider the following tradeoffs (all of which are from me, not my interlocutor): Cancelling a toll-access journal in order to fund author fees or to underwrite memberships with open-access publishers. Using staff and technology resources to collect student research in an institutional repository because it's what can be collected in the short term, even though doing so has no appreciable short-term impact on the access crisis.
Here's my answer: For how long? Short-term, of course toll-access access will suffer at the hands of any resource shift toward open access. But thinking short-term is exactly, exactly, what got us into the serials crisis to begin with. It's exactly what saddled us with the Big Deal.
We can keep feeding the same broken system in hopes it will become less broken. We can do this. It makes no sense to me, but we can. (What happens when it breaks down completely, which is not outside possibility? Good question. I've never yet seen a short-term thinker with an answer.) Or we can place some longer-term bets, with the explicit understanding that some of them will turn up losers.
I'd rather place the longer-term bets, myself.