Saturday, July 22, 2006

Economics of Open Access Publishing: another look


Analysis of data supplied by Magaly Báscones Dominguez in Serials earlier this year, supplemented by information from CERN, presents strong evidence for the affordability of one of the potential open access business models, OA by processing fees. This data suggest that OA by processing fees would be feasible, assuming reasonable but realistic processing fees, using library periodicals subscriptions as the only source of revenue. As an exceptionally research-intensive organization, CERN represents the worst-case scenario for this business model, which costs research organizations proportionately more than other kinds of organizations. Therefore, evidence of the affordability of this business model at CERN strongly suggests overall affordability of the model. Taking the CERN library periodicals budget as a base, an average per-article cost of 1,776 CHF / article (1,132 Euro, $1,436 USD) would be possible using current expenditure levels. If the IOP charge quoted at 573 Euro for New Journal of Physics were the average - CERN library could pay for a fully OA-by-processing fee model - AND, save half of its periodicals budget, too.

This is not an endorsement of the OA-by-processing fee approach, nor is it meant to suggest that all the revenue for open access should come from library subscriptions budgets. This is only meant as an illustration, that this is well within the realms of possibility.


According to Domingues, the CERN library periodicals budget is .669 million CHF. It is assumed that this sum would be supplemented by an equal contribution by non-CERN authors (1), for a total of 1.338 million CHF. The CERN list of publications for 2000 - 2004 lists a total of 3,766 articles published in peer-reviewed journals, an average of 753 articles per year.

1.338 million CHF (total of CERN library periodicals budget, adjusted to account for non-CERN co-authors), divided by 753 = 1,776 CHF / article (1,132 Euro, $1,436 USD).

In other words, if processing fee charges average 1,776 CHF / 1,132 Euro / $1,436 USD or less, switching from library subscriptions to libraries processing OA processing fees is possible, using current periodicals expenditures and no other source of revenue.

That this average is more than sufficient to cover quality publishing services is illustrated by the IOP processing fee charge quoted as 573 Euro for IOP's New Journal of Physics. Other OA publishers, such as BioMedCentral and Hindawi Publishing, are able to offer STM OA publishing services at below these averages (some Hindawi journals charge $495 US per article, for example).

If the IOP processing fee of 573 Euro were the average, a full OA-by-processing-fee model would save CERN half of its library periodicals budget. Another area of savings for CERN library would be interlibrary loans costs.

Based on

Magaly Báscones Dominguez'"The Economics of Open Access" Serials 19(1), March 2006, in which she claims that the processing fee approach to open access would cost CERN much more than current library subscriptions.

1. The estimate of the relative contributions of CERN and non-CERN authors is based on Table 5, page 6, which presents scenarios for OA by processing fee for Journal of High Energy Physics if costs are shared with non-CERN co-authors vs. CERN paying the full cost. The total for CERN paying the full cost is just over the twice the cost if sharing with non-CERN co-authors.

This appears to be a well-intentioned, but very biased, article. As evidence of bias, consider that the author, writing on open access, chose to publish results in one of the very few journals whose publisher has not given the green light for self-archiving, according to the Sherpa Romeo Publisher Copyright Policies and Self-Archiving, which lists no publisher information for Serials. In Table 5, p. 56, the author explores a potential processing-fee approach to Journal of High Energy Physics, and calculates differential fees based on whether CERN paid the full cost, or shared the cost with non-CERN co-authors.

Table 7, p. 58, and the author's conclusions, do not consider the co-author factor. The author's conclusions are based on journal-by-journal comparison, looking at select journals, which CERN authors publish in more frequently, apparently. Journals in which CERN authors publish in less frequently (CERN lists 174 different titles in which CERN authors published 10 articles or less from 2000 - 2004, for example), would obviously present a very different picture. That is, it is the journals CERN authors publish in most frequently for which processing-fee approaches would cost the most. If few articles are published in a journal, the processing fee will be low compared to the subscription cost. In other words, the author chose to present only data that would support an OA is expensive hypothesis, ignoring other data.

How these rather obvious biases got through a peer review process is another question.

For other criticisms of this approach, please see Peter Suber's Good facts, bad predictions in the June 2006 SPARC Open Access Newsletter.

This post reflects my personal opinion only and does not represent the opinions or policy of the BC Electronic Library Network or the Simon Fraser University Library.

No comments:

Post a Comment

Thank you for your comment. Comments on IJPE are moderated.