Both the cost of producing journals and the cost of acquiring them were topics of discussion. Participants discussed how journal costs could be kept low and how the high prices of journals are affecting colleges, universities, and industry. Several publishers talked about the challenges of pricing their journals and different models of financing journal publication.
THE PRICE TAG FOR BUYING JOURNALS
The cost of journals has risen sharply in the past years, and several participants had ideas of how to bring these costs down. “Solving the problems of expensive journals, and therefore inaccessible to some, will require a very concerted, courageous, and maybe even revolutionary effort by five groups of people,” Christopher Reed said. He listed professional societies, librarians, presidents and provosts, scientists, and funding agencies.
Professional societies should better serve the discipline by providing open access as soon as possible, without bank-rupting the organization, Reed said. The societies’ computers have become the libraries of the world; therefore the societies need the ethics of librarians, not of publishing houses, according to Reed.
Reed believes that libraries are wasting an enormous amount of money binding all copies of a journal. He believes that libraries are becoming white elephants and librarians should teach people how to use the electronic library. He called on presidents and provosts to cut library budgets. “Why are profit-making publishers attracted to this area? Because there is money—if you decrease the amount of money available, we will have less of this,” Reed explained.
Reed also said provosts and presidents have to install disincentives for promotion, to discourage faculty from sitting on the editorial advisory board of what he called a “junk journal.” Reed asked his fellow scientists to take a pledge not to submit to, referee for, subscribe to, or accept editorial appointments on expensive, non-open-access journals.
Gordon Hammes also asked participants not to support journals that have high costs per page, with the exception of thin journals in important fields that might have a higher cost basis. A number of participants echoed these thoughts. Steven Heller also suggested cutting sales and marketing staff to save money. However, funding agencies could also contribute to lowering journal prices, Reed pointed out, and he listed some ideas to cut costs. Funding agencies should take the publication business more seriously and bear some responsibility for it. There might be a need to explicitly deemphasize the number of publications at grant renewal time, fund for longer periods of time, discourage people from publishing little bits all the time—the least-publishable units—and shorten proposals so that reviewers actually have to read the papers.
The actual cost per user has declined for some journals. The cost per article for the average user has decreased, as a result of increasing expansion and increasing usage, from $12.00 an article in 2001, to $2.00 in 2004, Patrick Jackson said. “We hope to go to $1.00 next year,” Jackson continued. He said future chemists would demand that publishers continue to invest heavily in both innovation and quality. Elsevier has data that are currently stored in petabytes, which means the necessary infrastructure is extensive.
Subscription costs per page, however, vary from journal to journal. Hammes compared estimated costs per page for a number of journals and called for uniform page costs to reduce overall library spending. According to Hammes, the Journal of Biological Chemistry costs 4 cents per page; Proceedings of the National Academy of Sciences (PNAS), is 8 cents per page, and Biochemistry is 19 cents per page. Science does pretty well at 16 cents per page, Hammes said,
but most journals do not carry the same amount of advertising that Science does. Nature is a very expensive journal at 86 cents per page, according to Hammes. He said that switching to web-only journals would cut costs by about another 20 percent. Combining both of his suggestions—cutting some of the high-cost journals and going online—would leave libraries in pretty good shape.
EFFECTS OF PRICE INCREASES ON COLLEGES AND UNIVERSITIES
According to some participants, the increase in journal prices is forcing many schools to cut subscriptions, resulting in a lower quality of education. Between 1990 and 2000, the budget for science journals in the nation’s predominantly undergraduate colleges and universities rose by 120 percent, but this number is deceptive because the increase in journal prices over that period of time was even higher. This means that journals were being eliminated, Michael Doyle said. Martin Apple cited data from the Association of Research Libraries,1 which showed that journal prices have increased at a rate six times the Consumer Price Index during 1986-2000. Because library budgets are shrinking, and journal and monograph costs are soaring, increasing cancellations must follow, Apple continued. This trend will likely continue into the future as library budgets continue to shrink, he said. State universities especially face severe problems. An increase in future state and federal budgets for libraries, research, or other educational expenses should not be expected. “In fact, before the House and Senate right now is the first cut in the National Science Foundation, practically since it was founded,” Apple said.
Some publishers do offer discounts to schools. Martin Blume said that the American Physical Society offers online-only for 15 percent less for larger institutions and 20 percent less for smaller institutions. However, for some institutions, this may not be enough. Grace Baysinger said that many of the small schools could not afford Chemical Abstracts Service (CAS), even in consortium agreements, which might very well have an impact on tomorrow’s researchers. Baysinger reminded participants that 50 percent of Ph.D. students come from small schools, most of which have very limited access to CAS. The ability to find information is critical, and the ability of the discipline to support access to the major database is important, Baysinger pointed out.
Dennis Chamot drew a parallel to health care. Reed agreed, and said that a recent book suggests that 20 years ago, health care costs for administration were about 10 percent. After privatization, they are now about 17 percent. “So the profit motive adds to the cost of health care and doesn’t make it any more accessible,” according to Reed. He said that this is why he thought profit makers do not deserve to be in the business.
THE ROLE OF PROFESSIONAL SOCIETIES
Nicholas Cozzarelli said he did not think that librarians really want to support all of the activities of the American Chemical Society or the American Society of Biochemistry and Molecular Biology, but rather they want to buy that society’s journal. He thinks the system has brought about some excesses and listed the salaries of ACS officers as an illustration. Charles Casey spoke for himself (as the 2004 ACS president) and said that he was lobbying for more openness and internally trying to get some moderation in ACS salaries.
Apple pointed out that most societies in the Council of Scientific Society Presidents (CSSP) barely break even on their journals. The average loss per year is about $200,000, and they cannot shoulder additional burdens, he said. The science association publishers’ competitive advantage is higher quality. In cost per citation to the library, the difference may be ten- to fifteen-fold in purchasing from association journals versus commercially published journals, Apple said.
Michael Keller said that HighWire Press can help publishers, especially not-for-profit publishers, compete more effectively. HighWire serves 125 publishers, mostly from scientific societies and a few for-profit publishers. He said that HighWire Press was established to do two things: (1) to use network technology to enhance scholarly communication, and (2) to make not-for-profit society publishers more competitive.
Publishing is a very large commercial market; in 2002 it was about $7 billion according to data that Apple showed from the Electronic Publishing Services (EPS) Limited. “Because it is a big market, that makes a lot of difference to a lot of people about where it should go and what it should do,” Apple explained. Elsevier accounts for about a quarter of the entire market, he said. The top four providers account for about half of the market, and there are only a couple of major professional societies among the top 15—the American Chemical Society and the Institute of Electrical and Electronics Engineers (IEEE). He said it should be recognized that the societies are in the business of helping develop the next generation of scientists, as opposed to making profits for shareholders.
MODELS FOR FINANCING SCIENTIFIC JOURNALS
There are several different payment models for financing journal publication—library pays or subscription model, author pays, pay per view, and a combination of any of the above, possibly with page or color charges. Participants voiced their opinions on the strengths and weaknesses of
For more explanation of the data and more recent statistical analysis, see the Association of Research Libraries web site at www.arl.org.
each and discussed what a new model could mean for their institutions.
The most prevalent payment model is “library pays” with no extra page or color charges to the authors, Reed said. He did not agree with the common notion that the indirect funders—the government and research foundations—really pay for these costs. On the contrary, indirect costs for libraries are capped, he said. The institutions pay out of funds that could be going to research and education, according to Reed. However, subscriptions come with an extra cost for publishers, Cozzarelli said. PNAS loses about a third of all subscription revenue to subscription agents and other intermediaries, a fact which makes subscriptions financially inefficient. The subscriber model is market driven, Peter Gregory said. Gregory listed some of the consequences of subscription models such as centralized buying, a stress on library funding, and the expectation that industry pays it way. One of the strengths of the subscriber model in comparison to the “author-pays” model is that publication cannot be bought (i.e., as if it were an advertisement), Gregory said. A combination of library pays with additional page and color charges to the author constitutes the second payment possibility.
Pay per View
Pay per view is yet another model of financing journals. This model has its own pitfalls. Publishers might emphasize content that is likely to be highly viewed, and dull research might not be published, Gregory said. “And the consequence of that, again, would be that scientists and chemists especially would be repeating the dull but correct stuff again and again because no one bothers to publish it,” he said. The pay-per-view model is useful only as an additional service, Gregory concluded. He said it is far more important that the publication is based on the ability to do science rather than the ability to pay to publish the results.
Another model for financing journal publication is an up-front payment by authors, but problems arise with this model: wealthy research groups plead poverty and instead use their funds for other expenses. These charging models have failed in chemistry and physics in the past, Gregory said. “The authors don’t want to pay,” he said. There are also ethical problems with an “author-pays” model. “If there is money involved, either with referees or with authors who can pay to be published, it is more likely from our editorial experiences that there is corruption waiting just around the corner,” Gregory said. Shifting the cost from libraries to authors might make libraries obsolete, and industry would be the net beneficiary of this model, he said. This might mean a loss of income for some publishers. The Royal Society of Chemistry, for example, derives around 45 percent of its income from industry, which would have to be found from other sources if author pays were the only model. Authors themselves might also have to scramble for money if the payment model were to be changed, because Gregory doubted that enough academic funding is in place to support the system. Administrative and billing problems might also arise; publishers would be dealing with every author, not just with a few thousand customers. Problems could arise for both publishers and universities. Publishers do not have the means to collect outstanding author fee debt, and universities would have to deal with publishers from all over the world in all different currencies.
There are some problems with an author-pays model in the context of open-access publishing, Gregory added. He fully agrees with the idea of complete access to information, but cautioned that unclear financing could irreparably damage the great heritage of the American Chemical Society, the Royal Society of Chemistry, and chemistry itself over hundreds of years. Stevan Harnad objected to discussing “open access” in the context of publication financing. He said that no government is mandating an author-pays model, but added that the United States, the United Kingdom, Canada, Australia, India, Brazil, Norway, Denmark, and a few other countries are recommending open-access models—where articles are freely available to anyone on-line immediately upon publication—not author-pays models.
Gregory replied that the U.K. research council’s draft proposal includes a requirement for 2004 in which author-pays models must be are strongly considered by all research councils. The question was raised whether the British response was guided by the fact that the publication industry there is more dependent on overseas subscriptions perhaps than in other countries. According to Gregory, a good deal of the science and technical publishing industry is based in or has major units in the United Kingdom. A high percentage of the Royal Society of Chemistry’s revenues come from overseas—about 85 percent—representing a significant injection of cash for U.K. science. He said that this revenue benefits the UK trade balance, UK employment, chemical sciences worldwide (because the RSC acts worldwide), and is overall good for the UK contribution to developing and supporting the chemical sciences. “But the main winner is the chemical sciences wherever they are, as the money is spent on science rather than on commercial publisher shareholders,” Gregory added.
A number of participants related their experiences with authors’ not paying page charges. Cozzarelli said that this was not a problem at PNAS. On the other hand, ACS did have a problem before abandoning the author-pays model. Physical Review Letters has voluntary page charges, but other APS journals do not. Blume said that authors choose where to publish, and they will choose a journal without
charges if they can. However, he pointed out that asking an author to pay so that there will be open access might change this response. That is different from asking for page charges where there are controls, which was the situation in the past, Blume said.
Cozzarelli added that money had no influence on the papers he published. Gregory recounted having had money offered to him from a corporation that wanted to avoid patent issues to publish a paper. Hammes thought it appropriate that institutions pay a substantial part of the cost, because publication furthers the institution, whose business it is to get the research out there. Stephen Berry said that the reason a federal agency or the not-for-profits support research is because it will generate a public good. A public good is an item that is not diminished in value by use, he said. Scientific public goods are special because they increase in value with use. Agencies that support research should be prepared to pay for publication, he said. Berry called for an economic analysis of all the plausible modes of supporting journals to determine which have the lowest transaction costs, and where the largest fraction of the money is spent directly in supporting the publication. He said that page charges may look good, but they have large transaction costs because of the overhead steps from start to finish. Gregory said that researchers might not have a research grant. “How are you going to pay from your nonexistent research grant a publication fee?” he said.
According to Leah Solla, Cornell University investigated what would happen to its costs in an open-access environment.2 Calculations were based on how much the library spends on subscription costs and how many articles Cornell authors publish. The study found that Cornell spends $1,100 for every article published by a Cornell first author. If larger commercial publishers are removed from that equation—Elsevier, Wiley, and Springer, which make up about 25 percent of the articles and about three-quarters of expenditures—the cost drops to about $400 per article. Cornell did not research all author charges or all personal subscriptions of all of the researchers on campus, but the study still found the library-pays model to be more favorable for the university. “I don’t think that that would nearly add up to $3,000 or $4,000, the kinds of amounts that we have been hearing about today,” Solla said. Cornell would not easily adopt the author-pays model, she said. An author-pays model would shift costs to larger universities, but is not clear who would benefit from this shift. Reed cited the Journal of Financial Economics, which charges for submissions and then reimburses authors for accepted papers, as a pricing model.
Although a shift in payment models may mean larger contributions for some institutions, they might benefit from other aspects of the change. Undoubtedly some institutions will wind up having a slightly larger contribution, Vivian Siegel said. Yet in looking at the outcome, those institutions then also have access to large degrees of information that they might not otherwise have. Focusing entirely on the library budget is really missing the point of what a change is all about, she added.
An author-pays model might deter researchers working in the chemical industry, said Parry Norling, who spent 35 years at the DuPont Company. He said that the chemical industry is often reluctant to publish. While the patents that scientists develop are paid for by the company, page charges for research articles might have to be paid out of the scientists’ own pockets, unless they could argue that there is a real benefit to the company. He summed up his experience at DuPont by saying that page charges are a barrier to publishing itself, not only a factor in the choice of journal.
THE COST OF ARCHIVING
One of the costs that open access would incur is the cost of archiving. Hammes talked about models to pay for keeping an open archive. He said that raising the subscription price of a journal slightly would amortize the cost of archiving in a very short time. He urged societies that do not follow this model to make their archives free now and solve a lot of accessibility questions in the process. Andrea Twiss-Brooks said that such models would aid small institutions that may not be able to afford all of the on-line subscriptions for current periodicals. Patrick Jackson, however, explained that Elsevier could not give open access to its archives, because the company had invested $40 million in them and needed to earn back the investment.
Hammes said that the Journal of Biological Chemistry has been giving free access to its files after six months for several years and has not lost money. He added that the number of subscriptions has gone down, but in the same way they have for other journals, as purchasers eliminate duplicate subscriptions. The ACS archive is used heavily, but not nearly as heavily as current subscriptions are, according to Twiss-Brooks. She said that making the archives freely available would not impact current subscriptions. Some participants, however, felt that giving access too quickly might damage publishers. Blume said the 28 percent of APS subscriptions from the smallest institutions are the most vulnerable in the event APS makes everything open access very quickly. With them, there is a potential revenue loss of 30 percent, and possibly more. “That is why we are cautious about this,” Blume said. There is a possibility of granting open access for the entire archive now, if the subscribers sponsor it, continued to sponsor it, and agreed to increase their contributions in the future. Blume added that two APS journals are now open access. The first, Physical Review Special Topics—Accelerators and Beams, is supported by institutional sponsorship. Charging the author or the author’s
Available from the Cornell University institutional repository: http://techreports.library.cornell.edu:8081/Dienst/UI/1.0/Display/cul.lib/2004-3.
institution about $1,000 an article will support the upcoming Special Topics—Physics Education Research.
Jackson said that in general if a chemistry customer buys the back-files from before 1995, statistics indicate that the usage is usually around 15 percent, a number he thinks is significant.
AN INDUSTRY LIBRARIAN’S APPROACH TO ACQUIRING CONTENT
Lou Ann Di Nallo explained Bristol-Myers Squibb’s (BMS) approach to acquiring content. Her company licenses content globally so that it is available to its researchers no matter where they are, and trains its employees to use the electronic scientific literature to leverage its investment. Her company works through the purchasing organization Global Strategic Sourcing to obtain access to these journals though, which can be a challenge. “There is a little bit of a learning curve there for them to understand and accept that there are not four different companies out there waiting to sell you access to Tetrahedron, there is one and you have to deal with that,” she said.
According to Di Nallo, the BMS library’s strategy is often more of cost avoidance than cost savings, which has a lot to do with BMS being a very electronic environment. When, some years ago, there were significant budgetary pressures, the company canceled a lot of print, Di Nallo said. After an uproar by some researchers, some titles were reinstated, but after a year, statistics showed that print was not used heavily. The library keeps print from publishers who discount print with the electronic access. “The only reason we have the print is because it was actually at a discount,” Di Nallo said. She said the scientists are not really aware of the actual costs involved, just for the print. Added to this is the cost of having someone there to receive the print, to check it in, and to put it on the shelves, she said.
BMS continually reviews requests for new content through a couple of mechanisms. One is a user community of about 75 people representing different areas within the company. Then there is the Content Advisory Group, which is made up of senior-level members of the research institute. Around budget time, the company looks to this group to help it make difficult decisions. Di Nallo added that her company is almost obsessive about usage, statistics, and other metrics. Cost per use is a factor that guides budget decisions and negotiations with publishers. She said that this is not the only measure of value, but it is a pretty good one and something to which finance people can certainly relate. She said that the journals are expensive, but the cost of the journals pales in comparison to what BMS pays for the tools to access journal literature.
Di Nallo said that one of her challenges is to get the most out of her budget as content costs are rising. “There is a real balancing act that goes on there trying to figure out how to squeeze the most ‘bang for your buck’ out of the overall budget,” she said. Di Nallo said she felt that less competition in the publishing industry is leading to higher prices, and she may have to deal with this by limiting access. “We have been getting the word out there to our users about these journals. We have been putting linking solutions in place. We have been doing training education, and now we are really finding ourselves in the very awkward and uncomfortable place of having to think about limiting access to some of the stuff,” she said. She added that this has resulted in her library paying twice as much. Di Nallo also added that employee education is becoming a large component of her work, not only on what is available and how to use the tools, but also on the costs. “Typically, what I am finding is they are less inclined to pay for things than librarians are, once they are aware of the actual cost,” she said.
COST STRUCTURE OF STM JOURNALS
Editors from the Royal Society of Chemistry, the American Physical Society and the Public Library of Science, shared details and challenges of the cost structure of their publications with participants.
Royal Society of Chemistry
Several factors drive the cost of publishing a journal, Gregory said. Inflation, attrition, a publisher’s investment in its electronic platforms such as ScienceDirect, Wiley Interscience, or those of the ACS or the Royal Society of Chemistry, and the loss of subscriptions all contribute to rising prices for the Royal Society of Chemistry’s journals. Increased submissions and increased publishing output also drive the pricing considerations of all publishers, Gregory said. The number of the Royal Society of Chemistry’s core journal submissions increased by 17 percent from 2002 to 2003. This translates to 17 percent more work for the staff, or more staff, he said. “The question of the rejection rate came up early, high ones, low ones; who does [the review]?” he said. The Royal Society of Chemistry has 70 chemists dedicated to conducting the peer-review process. It rejects about 25 percent of submissions without even putting them out to other peers. The result of increased rejection rates for massively increased submissions is 6.5 percent more output from Royal Society journals in the period 2003-2004, he said. The European Community adds a 17.5 percent VAT (value-added tax) bill to electronic services, he said. As a consequence, the Royal Society of Chemistry supplies its customers with print, because many are actually buying print to avoid paying the VAT.
American Physical Society
Blume talked about the economics of American Physical Society publishing. As with the Royal Society of Chemistry, the number of submitted and published articles is the basic
factor that drives cost, coupled with inflation, Blume said. It often costs more to reject an article, in editor’s time, than to accept it, he said. Other costs include the office, telephone, postage, composition, production, and distribution. These costs are important, Blume said. Distribution costs are now for both electronic and print with electronic-only leading to significant savings. Active subscribers are about 20 percent electronic-only, not counting consortium agreements, which give electronic free distribution, he said. According to Blume, the net effect of dropping print varies is a cost reduction of 15 to 20 percent, or more in Europe, because subscribers not longer have to pay airfreight.
Blume compared the cost of the APS-all package—which encompasses Physical Review A, B, C, D, E, Physical Review Letters, and Reviews of Modern Physics—for a particular large institution in 1988 and 2004. In 1998, APS-all cost $12,015. In 2004, the cost increased to $24,570, with smaller institutions paying less than larger institutions due to newly introduced tiered pricing. However, this institution actually paid less to access the ACS-all package in 2004 than 1998, because in 1988 it held three APS-all subscriptions—one additional subscription to Physical Review Letters, and one additional subscription to Reviews of Modern Physics, whereas in 2004, the institute simply had one APS-all subscription. Factoring in the duplicate subscriptions, the total cost for the institute in 1998 was actually $38,300. In 1998, print was included with 90,000 pages; in 2004, it was print with 110,000 pages. “You would think that this would be enough to tell people to not just look at the prices, but look at what you get for it at the same time and what has happened in the interim. The large institutions have saved heavily,” he said. The cost per article for large institutions is 16 cents a page. APS tries to price its journals so that the cost per page is the same across all journals.
As for other publishers, rising submissions drive prices. Last year, APS received 27,000 submissions, which requires a significant editorial staff. APS has an in-house staff of 35 editors plus about 50 editors based around the world—so-called remote editors like Jack Sandweiss at Yale, the editor-in-chief for Physical Review Letters. There is a widening gap between the number of published articles and the number of submitted articles, Blume said. He said that the average cost of an article is $2,000. Eliminating print would lower this even more. Blume said that APS is lowering prices next year because of the benefits from electronic work in the office.
Public Library of Science
Siegel talked about the Public Library of Science (PLoS) and its business model. This model involves recovering all of the costs of publication (including peer review, production through the on-line version, and all associated overhead costs) through up-front publication charges. Up-front charges are currently set at $1500 per submission.
According to Siegel, PLoS thinks that up-front charges should cover publication through the on-line version, and it then sells print versions of its journals at the cost of printing and circulation, supplemented by print advertising. When a researcher submits an article, he or she indicates what part of the charge the research can pay. This information is shielded from anyone who can make a decision about what to publish. Currently, slightly less than 5 percent of the total publication revenue is lost to nonpaying or partially paying authors, Siegel noted. There are no additional charges, no color charges, and no individual page charges. There is no arbitrary size limit to any of the papers. PLoS has several grants through the startup period. It also has membership programs, and allows corporate sponsors, albeit very carefully, Siegel said.