Although the authors don’t say so directly, nationally – as a table in the study shows – non-mandated benefits are about 30% of total comp for state and local workers. (Mandated or legally-required benefits are things like Social Security which are roughly the same in public and private jobs.) So to close the 6% gap, benefits must be approximately 20% higher in the public sector compared with the private (6/30 = .20).
Undoubtedly, critics will note that the authors evaluate the benefits at current employer costs. So if there are unfunded promises being made (pensions & retiree health), it could be argued that their estimate of benefit value is too low. But note that even if benefits are 50% underestimated (i.e., if the benefit premium were 40% instead of 20%, you would end up with overpayment in the public sector by only 6%.
The study is at http://www.irle.berkeley.edu/cwed/wp/2010-03.pdf
A press release from UC on the study is at http://www.universityofcalifornia.edu/news/article/24306
Finally, a news account is at http://www.sfgate.com/cgi-bin/article.cgi?f=/c/a/2010/10/19/MNUJ1FUAOH.DTL&tsp=1
No, the authors do not break out UC employees and it probably could not be done using the database they had.