Thinking more about how book processing charges concentrate costs.
The largest four monograph publishers in the UK (CUP, OUP, T&F, Palgrave) published 5,023 monographs in 2013 (source: Crossick report). At a £6,500 BPC (CUP price) this would cost £32,649,500. At an £11,000 BPC (Palgrave price) this would cost £55,253,000. At Ubiquity Press’s BPC of £5,050 (including copyediting) this would cost £25,366,150.
In a national context, ~27,000 books were submitted to REF 2014 so far as I can see (source: REF2 items with an ISBN). This is ~5,400 UK books per year (divided by five [years]), so higher than the above estimates. The total UK library spend on purchasing books (of all kinds not just monographs) in 2010/2011 was ~£60m (£47m on print, £13m on ebooks) (source: SCONUL).
Because the UK produces more books than it purchases, a gold OA transition is hard work on current purchasing budgets in this national context, despite funders stepping up to provide this in some cases. While on some levels of fee it is possible, it is also the case that even if we switched to 100% gold OA monographs funded by BPCs tomorrow, UK libraries will continue to have to purchase books from abroad. So offsetting only occurs if an immediate international switch happens, which it won’t. It is additionally the case that the print costs don’t go away, because researchers still want this to read 80,000+ words. There are also problems of selection and purchasing that make offsetting far harder in a books environment. Plus, these funds are not well distributed; some institutions have nothing. Likewise, independent scholars have no recourse. It would be interesting to see what percentage of the purchasing budget goes at the moment to paying for books by UK researchers vs. those abroad.
Edit 2015-11-18: I accidentally shaved £10m off the SCONUL budget. This has now been fixed.