Is free affordable?

The Journal of Clinical Investigation (JCI) has been providing free and unrestricted access to its content ever since it went online in 1996. We have been able to support the costs of this online Open Access through a mix of page charges and income from subscriptions to the print version of JCI, which itself was first published in 1924.

Why do we favour online Open Access? Not only has it afforded the journal an opportunity for exposure to a wider audience than it might have reached with the print version alone but we have also found that it is a policy that the journal can afford. The larger goal of preserving free access indefinitely was perhaps not explicitly appreciated nor stated during the years of transition to the online realm, but it is now. Open Access has been, and remains, the right thing to do.

However, the main problem with maintaining a free online model is that it adds to the expenses of the journal, which are considerable for the JCI. In 2003, the total cost of publication was about $2,580,000. Producing, manufacturing and distributing the printed version are a significant, but not the biggest, part of that cost. Other expenses comprise, in descending order: staff salaries and benefits; compensation to members of the editorial board; delivery and hosting of the online journal; and general and administrative costs, such as travel and rental of office space.

The JCI of course generates revenue, and even a modest surplus. In 2003, the journal earned around $2,840,000 in revenue. Print subscriptions accounted for more than $1.5 million of the total; submission and publication fees, the next largest revenue block, were a little more than $950,000. The remainder came from reprints and royalties.

With the online version free, the fall in subscriptions to the print edition is, not surprisingly, a concern. I believe, however, that there is little most publishers can do about the long-term decline of institutional subscribers,such as libraries, the revenue from which constitutes most of total subscription income. In 1993, JCI fulfilled almost 3,600 institutional subscriptions, but by 2003 this number had fallen to 2,001 (Figure 1). The journal competes with others for a portion of already tight -- and tightening -- library budgets, and it competes with itself by offering, for free, not only a replica of its printed version online, but a version rich with capability for multimedia supplements, interlinking and interoperability.

There is little complexity, at least from a budgetary standpoint, in the decision a library faces between paying for the print version and not paying for what is, technically, the better online version. That noted, the decline demonstrated in Figure. 1 is not as drastic as it perhaps might be. While it cannot be known with certainty, the appreciation for print as an archival medium has probably prevented more rapid attrition. However, as methods develop to ensure stable, reproducible archives for online content, this attraction to print among institutional subscribers will presumably disappear.

At the same time as the number of institutional subscribers to the printed journal has eroded, the JCI's online readership has continued to grow. At the end of 1997, the JCI's online version was receiving visits from about 3,000 unique hosts monthly. At the end of 2003, this number had grown to about 55,000.

Viewed in ethical terms, anyone who wants to read a JCI article online can. Viewed in financial terms, the cost to provide access to an additional host is far less than the cost to provide a printed copy to an additional reader. However, the basic costs of operating and producing the JCI remain. In 2003, the JCI received 3,045 original submissions, 695 revisions, and uncounted e-mails, letters and phone calls, all requiring scrutiny by a compensated editorial board -- with the notable input of unpaid consulting editors and reviewers -- who provided 5,045 reviews, also requiring evaluation by the editorial board -- using a technical infrastructure that requires funds to maintain and develop.

The JCI published 308 research articles in 2003 (about 90% of submissions were rejected). Upon acceptance, an article is edited for accuracy and conformity to the journal's editorial style; its figures are revised according to the JCI's design aesthetic; the components are flowed into a page design and massaged through a variety of proofing steps before it's ready for publication, whether online or print.

For the print version, there are additional proofing steps before the presses run. For the online version, an article is broken apart into its components (text, tables, figures, supplementary material) and processed into formats required for online presentation; this processing entails yet more proofing. In addition, there are non-trivial costs associated with managing the delivery, storage and development of both versions.

Publication fees (or page/author charges) have traditionally been employed by many publishers to defray a portion of the expenses related to the entire publishing process. (It should be noted that the JCI discounts publication fees for authors who are unable to pay them, as is the case for many other publications.) In the case of the JCI, publication fees have been a longstanding part of the journal's business model. In 1993, publication fees represented about 25% of that year's total revenue of about $2 million; in 2003, these fees were almost 35% of the revenue total.

The crucial question regarding publication fees in relation to Open Access is not what percentage of the revenue whole they constitute, but what percentage of the expense as a whole they will have to be to continue to support publication. The answer --without muddying the waters with prospects for grant support and the like -- is 100%.

It is an interesting exercise to interpret the finances of the JCI for the last several years through an Open Access lens: extracting, as reasonably as possible, the costs associated with production and distribution of the JCI's print version, and then dividing the remaining expense of publishing by the number of research articles in each year. The resulting hypothetical average publication fees per article from 1996 through 2003 are presented in Figure. 2. The fee that accrues on average to a JCI article at present does not reach the amount noted in Figure. 2, but neither is the fee a small fraction of the hypothetical. Publication fees, at least in the case of the JCI, constitute a meaningful portion of the revenue supporting the journal.

The question arises, what type of journal publishing-in quality or quantity-can publication fees support? Happily, the average publication-fee burden per article is modified depending on the number of articles a journal is willing to publish. However, increasing the number of articles leads to reduced impact (or a reduction in top-tier/premier status), which in turn reduces the quality of submissions from which a journal is able to choose, which further depresses impact. This affects a journal's ability to attract editorial expertise and to solicit the opinions of unpaid reviewers who have many other demands on their time, and who might prefer to provide their opinions to journals of higher quality.

One might whittle away at the publishing process as a whole; there probably are values added that could be subtracted without entirely corrupting the process. Perhaps more of the process can be left to automation. But the alternative to a publisher producing something that the research community continues to recognize as a useful publication is a declaration of bankruptcy.

The JCI has the benefit of history; it has the benefit of being supported in many ways by authors and readers; it has the benefit of having a blend of income sources. Increasingly, however, it will not have the benefit of the level of subscription income on which in part it currently relies. This translates into further reliance on publication fees. It seems fair to claim that it is unclear whether the system as a whole can adapt to support the quantity and quality of Open-Access publications envisioned by Open-Access advocates. And I am among them.

John B. Hawley
Publisher, The Journal of Clinical Investigation, The American Society for Clinical Investigation

The author is grateful to Ushma Savla, Executive Editor of the JCI, and to Karen Kosht, Managing Director of the ASCI, for reviewing and commenting on earlier drafts of this article.