I really can't explain why, but PLOS tries to: https://www.plos.org/publications/publication-fees/

They just increased the fees for their general purpose publication, PLOS ONE. Here's their reasoning:

"PLOS ONE has not increased its Article Processing Charge (APC) since 2009, for the past six years absorbing increasing publishing costs without raising author fees. Its new price reflects the work involved in shepherding the volume of papers PLOS manages from submission to published work; PLOS invests significant resources to improve the quality of PLOS ONE output, thoroughly checking for ethics, competing interests and robust science."

They also say they're updating their terrible article submission system. I don't see how any of this costs as much as it does. PLOS ONE alone publishes ~30,000 articles a year, that's a huge amount of money in fees, even if not everyone pays. On top of all of that, they accept and publish 70% of all articles. That's a really leaky sieve that shouldn't be so expensive to maintain.

I don't know why PLOS charges more. The point I'm trying to make is that - just because there's an OA journal out there which charges $100 per article; it doesn't mean that the costs are justified.

Do you have any compelling evidence that the costs for OA journals is one or two orders of magnitude less than everyone claims they are? (See e.g. http://www.nature.com/news/open-access-the-true-cost-of-scie... which also shows a wide variety in publishing costs.)