Jim Till's post on Open Access Science and Science Policy on Be Openly Accessible or Be Obscure, nicely sums up one of the most important reasons why we need open access.
The most rapid advances in science come with open sharing of information, and collaboration. That is how the world's scientists accomplished the mapping of a human genome in a matter of years. If traditional publishing practices had been followed instead of open sharing, it seems likely that mapping the human genome would have taken decades, if not centuries.
Our world shares some issues on a global basis; some of these are, or will become, urgent. One example is global warming; surely this needs the kind of open sharing and focus on the problem that went into the human genome project.
Another example is bird flu. The more our neighbours know about viruses, the better equipped they are for early identification and dealing with an epidemic, the lesser the chances that the epidemic will arrive at our shores.
We will save money with an open scholarly communications system, as preventing people from reading has significant costs. However, even if it did cost more, the question would not be how could we afford OA, but rather how could we not.