Here is a 2010 research paper (pdf) by Schenk and Matsuyama, and here is one sentence from the abstract:
Our results show returns are six percent for those completing a community college degree.
Other work shows that fewer than forty percent of those who start will finish with a degree-based reward and that number may be lower yet for the marginals who are not currently enrolled. Of course those who do not complete will waste some time and money along the way, although the learning may raise their productivity somewhat.
The overall return on additional subsidies to community college participation is therefore…?
The initial pointer is from @herdingbats.