It's a complex problem to be sure, and not entirely mismanagement. In some cases it is, like the links I posted. E.g:
-Where one successful project is put on hold so the scientist can prove to management the dud is really a dud.
-Or another where his project was simultaneously outsourced to India, which failed. But it made management look bad, so they wrote it off his department's expenses. When their formulation worked, it was rejected because it was over-budget per/pill (but only because the India outsourcing was added to their tab), otherwise they would have been fine.
Other times it's just the pharma ecosystem - I wrote about this a few weeks ago: http://news.ycombinator.com/item?id=3376428
I think the big problem with exploiting existing revenue streams is that A) most of the blockbusters developed last decade are going off patent very soon, with nothing to replace them and B) you can only develop so many "reformulations" before people realize what is going on.
Pharma is going to hit a point in the very near future where their revenue drops by multiple-billions. They are scared silly over this (rightfully so). I see two options:
1) Actually start doing science again, instead of firing scientists and reformulating the same pills over and over. This isn't going to happen and the time to do it was ten years ago. They are way behind the curve now.
2) Embrace the fact they are glorified marketing firms, fire all their scientists and use their accumulated wealth to purchase drugs from small companies, then push them through clinical trials. They also need to stop thinking about blockbuster hits - those days are over. This is basically what you are suggesting.
Realistically, Pharma is moving towards option 2. But they are moving towards it so slowly that they are cannibalizing themselves along the way. It is never pleasant to realize your model is obsolete, but the days of Big Pharma and billion dollar blockbusters is (probably) over.