Yale science Libraries have stopped paying the article processing charges for Yale faculty who publish in BioMed Central journals. Yale says:

Starting with 2005, BioMed Central article charges cost the libraries $4,658, comparable to single biomedicine journal subscription. The cost of article charges for 2006 then jumped to $31,625. The article charges have continued to soar in 2007 with the libraries charged $29,635 through June2007, with $34,965 in potential additional article charges in submission.

BMC responds:

The main concern expressed in the library's announcement is that the amount payable to cover the cost of publications by Yale researchers in BioMed Central's journals has increased significantly, year on year. Looking at the rapid growth of BioMed Central's journals, it is not difficult to see why that is the case. BioMed Central's success means that more and more researchers (from Yale and elsewhere) are submitting to our journals each year. [...]
An increase in the number of open access articles being submitted and going onto be published does lead to an increase in the total cost of the open access publishing service provided by BioMed Central, but the cost per article published in BioMed Central's journals represents excellent value compared to other publishers.

The increased cost arises because Yale researchers are submitting more and more work to BMC journals.  More manuscripts = higher costs, but if the cost per article has not gone up, then BMC's model scales effectively.  Here are some other ways to look at the numbers:

• For around $65K, Yale gets about 40 articles published OA, that is,available free to everyone everywhere forever, plus a "subscription" (that is, Open Access, like everyone else) to 179 journals.  Theaverage biomed journal subscription is around $1000-1500/yr; choosing the lower figure to be conservative, those subscription-equivalents are worth $179K/yr.  Even if Yale only wanted to subscribe to around a third of the BMC journals, that would still cost about the same as the OA charges --and this comparison ignores the page, color and miscellaneous charges that many journals levy.  (An example: PNAS charges $70 per printed page, plus $325 for each color figure or table; $150 for each replacement or deletion of a color figure or table.)


• Yale could publish those 40 articles elsewhere without paying anything (again, ignoring page etc. charges).  Assuming they don't subscribe to any of the journals they publish in, though, every time any Yale employee wants to read one of those articles they're on the hook for somewhere around $30; so it only takes 2166 person-articles, or an average of about 50 employees wanting to read each article, to get back to $65K -- without the benefit of OA.


• Yale spent about $7.7 million on subscriptions in 2005-6; converted to OA author-side charges at $1600/article, that's about 4800 articles.  A PubMed search on "Yale" gives 2272 hits; "Yale in title/abstract" gives 131, leaving 2141 papers where "Yale" is probably in an author's address.  I can't find a quick way to break out Yale's subscription expenditure by field, so what proportion of the $7.7mil goes to biomed journals I couldn't say (though STM titles are the most expensive subscriptions for any academic library). If PubMed-indexed journals make up 44% (2141/4800) of Yale's subscription costs, which does not seem unlikely, then they're already paying $1600/article -- without the benefit of OA.

A quick fiddle with biology + medicine data from theJournal Cost-Effectiveness database gives an average price per article of around $12 for toll-access journals, but that's (one subscription)/(total no. articles).  The question is, how many subscriptions do they sell -- that is, what is their income/article?  We know what BMC makes per article: about $1600 on average.  If an average toll-access journal sells just 135 subscriptions per year, they're bringing in more per article than BMC.

There's more, but that'll do for now.  Two questions arising:

1. what's the average page/colour/misc charge levied by toll-access journals?
2. how many subscriptions does an average journal sell each year?

---

An appendix of sorts: the BMC cost structure

Article Processing Charges

standard charge = $1600 (129 journals)

Supporter's Membership

Supporter Members pay a flat rate annual Membership fee based on the number of biology, chemistry, physics and medical researchers and graduate students at the institution. Members of the institution are then given a 15% discount on the APC when publishing in our journals.

So if this fee is to be less than 15% of total APC, total APC must be at least the figure in column 3.  Since the average is likely to be close to $1600/article, dividing through gives the number of articles in column 4.

Postpay Membership

...group members are invoiced in arrears for articles authored by their members that have published in our journals since the last invoice date. Invoice schedules are set on a monthly or quarterly cycle.

Prepay Membership

...enables an organization to cover the whole cost of publishing for their investigators when publishing in our open access journals. No additional fees will be paid by individual authors. This is an advance payment system whereby customers pay upfront for accepted articles authored by their investigators to be processed and published. Upon publication, the full Article-Processing-Charge (APC) for the journal in question, minus a loyalty discount, will be deducted from the account.

The higher the amount paid in advance, the greater the loyalty discount given on each APC.

No numbers seem to be available for the "loyalty discount".