The Cost of Publishing an Electronic Journal
A general model and a case study
Tilburg University Library
In their excellent review of the development of scholarly publishing in the United States, Tenopir and King  present evidence showing that the average institutional price of a scholarly journal subscription has increased from $39 in 1975 to $284 in 1995, a factor of 7.3 in just twenty years. Based on these figures, Tenopir and King conclude that: "It is clear that traditional scholarly publishing is in serious economic difficulty".
General inflation and increase in size (more pages per issue, more issues per volume, more volumes per year) of the journals, account for only 52 percent of the price increase. Tenopir and King explain the remaining 48 percent by pointing to the dramatic decrease in personal subscriptions which started in the late seventies. Publishers have apparently addressed this fall in revenue by increasing institutional subscription rates, thereby causing a vicious circle of cancellations and further increases in institutional rates.
This phenomenon, also known as the "serials pricing crisis" in the library community, has, indeed, advanced to a stage where one might seriously doubt the economic viability of the present printed system of scholarly communication.
Electronic journals, as distinct from electronic versions of printed journals, seem to offer at least a partial solution. Stevan Harnad [Harnad, de Kemp, 1997], founder of Psycoloquy, one of the first electronic journals, claims that electronic publishing leads to a cost saving of 70 percent, since in a purely electronic environment, only two cost categories remain: peer review and editing. Publishers, on the other hand, contend that these costs savings are highly illusory.
Arnoud de Kemp [Harnad, de Kemp, 1997], for instance, claims that peer review and editing are costly processes which are carried out outside the academic community, and there are also administrative and marketing costs to be taken into account. Marks [in Duranceau], director of the publications division of the American Chemical Society, estimates first copy costs -- acquisition, peer review, editing, conversion to digital format -- to account for more than 80 percent of the total publishing costs. These costs would also exist in a purely electronic mode of publishing.
Odlyzko , another prominent proponent of electronic publishing by the academic community, reminds us of the familiar 80/20 rule: the last 20 percent of a product or service accounts for 80 percent of the costs, "by throwing out a few features, it is possible to lower costs dramatically." He stresses the point, together with Harnad, that electronic publishing is essentially different from the print-on-paper paradigm, and that whereas the costs of printing will only go up, prices in information technology are dropping [Odlyzko, 1995].
To contribute to this ongoing debate, one of the goals of the project "Electronic Journal of Comparative Law" (EJCL), was to make an assessment of the costs of developing and running an electronic journal. This assessment, carried out by PricewaterhouseCoopers, is described in this article. The exercise was carried out in two steps. First, a general costing model was developed with which the shared and direct costs of producing an electronic journal can be charted. In the next step, actual data gathered from the experience with developing and publishing EJCL were used to illustrate the application of this model and to arrive at an assessment of the costs of publishing this particular electronic journal in an academic environment.
It is the intention that the general model, illustrated with real-life data, can help others to determine if a venture in electronic publishing is a feasible option. It should also spice the ongoing debate between publishers and academics about the true costs of electronic publishing.
This article is organized as follows. The next section presents a short description of the EJCL project, providing necessary background. Following that, the general costing model is presented. Next the model is applied to the EJCL situation. The article ends with some conclusions.
The EJCL Project
The EJCL project, carried out by library and computer center staff and legal scholars from Tilburg University and Utrecht University (Netherlands) set out to achieve the following goal:
- Developing an editorial, technical and organizational concept for an electronic review of comparative law which publishes articles of high quality in the English language contributed by Dutch and foreign scholars, as well as producing two issues of this journal.
The project ran from March 1997 till June 1998 with a grant from IWI, the Dutch platform for innovation of scholarly information supply. Both the controlling document as well as the final report of the project are available on the Web, for those who wish more information than given here.
The project was divided into seven so-called "work packages". Work packages one and two were surveys of existing print journals in comparative law and existing electronic journals in law, respectively. The results of these work packages served as input for the core work packages: three, which concerned developing the editorial concept for the journal; and four, the technical realization of the journal's WWW-environment, as well as implementing tools to support editorial workflow. Work package five covered the actual production, while the final two work packages were of an auxiliary nature. In six, the issue of costs was studied, of which this article is a reflection. Work package six also studied technical possibilities for differential access between academic and non-academic communities. Finally, work package seven was concerned with dissemination of the results of the project.
At the time of writing of this article, three issues of EJCL have been published, containing six extensive articles. The international editorial board, with Dutch, German, Swedish and British legal scholars, plans to publish four issues per year. With a grant from the Tilburg Faculty of Law, an assistant editor has been appointed to help the editorial board with the technical intricacies of electronic publishing on the Web.
It should be stressed that the journal's site is a simple one, which is primarily designed for distributing mainly textual material. Authors submit their manuscripts as either Word or WordPerfect files. After a reviewing process, these manuscripts are copy edited by the author in cooperation with the assistant editor. Finally, from the WordPerfect version, HTML, text and Word versions are produced, and the article is published in the available four formats on the journals site. Subscribers to the journal's listserver are alerted by electronic mail of newly published issues. The listserver used is provided by Surfnet, the Dutch academic network, and is one illustration of the preference for public domain solutions throughout the project.
A General Costing Model for Electronic Publishing
This section and the next are based on the accounting audit carried out by PricewaterhouseCoopers, which is also available on the Web.
The assessment of the costs involved in developing and running an electronic journal was done in two steps. First, a general Costing Model was developed with which the dedicated and indirect costs of producing an electronic journal can be charted. In this section the starting points of this general Costing Model are formulated.
Costs that were included in the Costing Model
The cost of publishing an electronic journal can be divided into the following categories:
- General overhead, e.g., buildings and management.
- Facilities, e.g., computers and networks.
- Publications, e.g., editing, training, and marketing.
- Creation of material, e.g., authors, editors and reviewers.
- User costs, e.g., personal computers and Internet connections.
The model concentrates on the costs of facilities and publications. The costs of authors, editors and referees are ignored, since they do not ask for a financial reward; this situation also applies for (most) traditional printed journals. We recognize the fact that some costs are transferred to the users of the journal (computer, software, Internet accounts), but they are left out of the analysis, since users already have the necessary equipment. General overhead is assumed to be the same for electronic and traditional printed journals.
Direct and shared costs
The model distinguishes between direct costs, which are assigned entirely to the cost of the journal, and shared costs, where only part of the costs is assigned. In order to separate the direct and shared costs of producing an electronic journal, one has to take into consideration the environment in which the journal is being produced. Two situations can be distinguished:
- Situation A: the project of publishing the electronic journal is based on a specific set of activities in a surrounding environment of facilities,
- Situation B: the project of publishing the electronic journal is based on a specific set of activities without a surrounding environment of facilities.
Considering these two situations, the following underlying questions have to be answered while determining the total project costs:
- Does the project make use of shared resources of the surrounding environment of facilities, which lead to indirect costs?
- Are there any costs that are directly assigned to the project?
If an electronic journal is published in a surrounding environment of facilities (i.e., Situation A), part of the costs involved have to be assigned to the project according to the specific use of resources for the project in relation to the total usage of these resources by the organization.
This principle applies not only to existing facilities used for the project, but also to newly made costs for the project, which are used outside the scope of the project. Besides these shared costs, the remaining costs involved will have to be assigned directly to the project, as they have no use for any other activities within the organization.
If an electronic journal is published without a surrounding environment of resources (Situation B), all costs will have to be assigned directly to the project.
Assignment of costs
With respect to the model for determining the costs of an electronic journal, different categories are defined to be able to declare the different costs. Costs were divided into certain categories by means of a four-step process:
- In the first step, the required activities within the project at a general level are declared.
- Next, these activities are grouped according to possible (most likely) base of calculation of usage.
- Then, it is determined whether the costs of these activities are shared with other activities, or whether the costs are directly assigned to the project.
- Finally, it is determined in which category the costs belong.
The outcomes of the first two steps of this process, the categorization of the costs, are as follows:
- local network: hardware, software, cabling, maintenance
- Internet (connection): hardware, software, maintenance, subscription costs
- servers: hardware, software, maintenance
- clients: hardware, software, maintenance
- web site: development, design, programming, testing, installation
- maintaining the electronic journal, including management, editing, and text and graphics
- instructing, training and supporting the users, including documentation on and procedures for the use of the electronic journal
- commercial activities, including marketing the electronic journal
In order to calculate the costs of the journal, a portion of each shared cost must be assigned for each of the cost categories. This is called a "declaration base". The declaration base is one of the variables within the model which can cause a lot of debate. In order to reduce this discussion, the variables are kept as simple and straightforward as possible.
The declaration base is determined by taking into account the nature of the majority of activities/costs involved in each category. The following units were used to measure the costs of shared facilities.
- network: volume of data transfer
- Internet connection: volume of data transfer
- server: storage
- client: storage
- website: staff time
Publications costs were all measured as staff time.
By simplifying the process further, three main variables remain:
- All costs involved in transferring data between servers and clients should be calculated in percentages of total transfer. (As there are no data available on the transfer usage for Tilburg University, a different method will be applied.)
- The activities of both servers and clients require the storage of data; therefore, the declaration base is a percentage of total storage.
- The costs for the development and exploitation of the web site and the organization are calculated on the basis of time spent on the project. The unit of measurement is full time equivalents (fte), which is to be multiplied by the average cost of an employee per year.
Shared costs can now be calculated by determining the total project usage for a specified cost category, which is then multiplied by the calculated rate for indirect costs for this category. The rate for shared costs for a certain category can be calculated by dividing the total related costs (including project costs) by the total related usage (including project usage).
The costs which are made solely for the project can be directly assigned to the Costing Model. For example, in case of the EJCL, costs for the development of the web site, organization and miscellaneous costs are considered to be specific for the development and exploitation of the journal and therefore directly allocated to the project.
A number of special factors apply in this model. In the style of electronic journal studied, �users� are subscribers to the journal and they are expected to download the journal once per issue. This is the reason for using storage as the declaration base for client costs.
Other assumptions that influence the calculation of the costs of the project, are:
- The costs of the project are calculated for a period of one year; investment costs are calculated for one year using a four year depreciation period; and costs of running the journal are defined for one year.
- The only costs taken into account are the costs that are made internally.
- The only medium considered for distributing the journal to external users is the Internet.
- The section "clients" in the model refers only to internal users, i.e., staff and students of Tilburg University.
Costing Model for publishing an electronic journal
The Cost Model can be summarized as follows. The costs taken into account are divided into direct and shared costs.
The costs, exclusively made for the project, were directly assigned to the Costing Model of publishing an electronic journal. These cost can be categorized into costs for the:
- local network,
- Internet connection,
- web site.
The calculation of the shared costs for the facilities is (per category):
assigned cost = total cost * fraction of service assigned to the electronic journalnetwork:
total indirect costs = actual transfer network * calculated network rate for the indirect costs
total indirect costs = actual transfer Internet connection * calculated Internet connection rate for the indirect costs
total indirect costs = actual storage servers * calculated servers rate for the indirect costs
total indirect costs = actual storage client * calculated clients rate for the indirect costs
Application of the Costing Model to the EJCL
To apply the costing model to the EJCL, some specific considerations and starting points have to be recognized. Of the two situations described before, the EJCL-project is based on a specific set of activities in a surrounding environment of facilities. The surrounding academic environment is Tilburg University with its existing infrastructure and facilities.
The project is based upon shared resources as well as costs that are solely assigned to the EJCL. The part of the costs that are assigned to the project are the dedicated costs and the shared costs that are based on available resources of the university.
The starting points of the EJCL are:
- Tilburg University has a contract with Surfnet, the organization which runs the Dutch academic network, for the Internet connection subscription costs for servers and clients of Tilburg University.
- Users/readers of the journal have the opportunity to choose from four formats (HTML, ASCII, MS Word and WordPerfect) when they download the EJCL. We estimate that readers download the journal once per issue.
- Printing costs are considered to be irrelevant for the assignment as they are approximately the same as the costs for making copies of a conventional journal. If readers in the library choose to read an article, they usually make a photocopy since journal issues can not be borrowed. Also, the copy or print is suited for annotation and highlighting.
The following facilities costs for the EJCL are shared costs:
- network hardware costs,
- network cabling costs,
- Internet connection subscription costs,
- server hardware costs,
- server maintenance costs,
- client hardware costs,
- client software costs (license costs).
Since there are no data available for transfer capacity nor actual transfer, it is impossible to compute the calculated rates for the network and Internet connection (a, b and c). To circumvent this difficulty, three alternative scenarios have been worked out to compute a different general base for mapping these costs. Scenarios 2 and 3 also apply to cost categories d, e, f, and g.
Scenario 1 - Time usage/capacity
- Tilburg University has 2,500 PCs available,
- usage of these PCs is 8 hours / day, 250 days / year
- (total estimated capacity of PC usage is 250,000,000 hours a year),
- 600 users / readers of the EJCL,
- 4 issues a year,
- 2 hours per issue usage of the PC,
- (total estimated PC usage for the EJCL is approximately 5,000 hours a year)
From these estimates a rate of 5,000/250,000,000 would follow, which is 0.02%.
Scenario 2 - Storage usage/capacity clients
- Tilburg University has 2,500 PCs available,
- average storage capacity of these PCs is 600 Megabytes (Mb)
- (total estimated capacity of PC storage is 1,500,000 Mb),
- 600 users / readers of the EJCL,
- 4 issues per year,
- estimated storage size per issue is 1 Mb (4 different formats),
- (total estimated PC storage for the EJCL is approximately 600 Mb).
From these estimates a rate of 600/1,500,000 would follow, which is 0.04%.
Scenario 3 - Storage usage/capacity servers
- storage capacity servers is 200 Gigabyte (Gb),
- estimated capacity in use is 70 %,
- (total estimated server storage capacity in use is 140 Gb)
- 4 issues a year,
- estimated storage size per issue is 1 Mb (4 different formats)
- (total estimated server storage for the EJCL is approximately 4 Mb).
From these estimates a rate of 4/140,000 would follow, which is 0.003%.
Analyzing the outcome of the different scenarios, we can draw the conclusion that the rate to be used will be relatively low. Taking into account the worst-case scenario which has been calculated, the total costs to be assigned to the project for network and the Internet will be less than (Dutch guilders) NGL 1,000,-- as the total costs of network and Internet together are approximately NGL 5,000,000,-- per year.
Thus, the shared facilities costs assigned for the EJCL are so small that they can be treated as zero in the Costing Model.
The following direct costs are assigned to the Costing Model:
- web site development costs,
- web site design costs,
- web site programming costs,
- organization maintenance electronic journal costs,
- miscellaneous project costs.
The table below gives the actual direct costs incurred in the project. It is based on the financial statement which was sent to IWI after the project's completion. For convenience, amounts are given in Dutch guilders (NGL), dollars and hours claimed by project staff. The hourly rate was NGL 84.375, the conversion rate for dollar to Dutch guilder used is 0.5 (1$ = 2 NGL), which is approximately the actual rate at the time of writing, early September 1998.
EJCL: Actual Project Costs
inv. print journals
inv. electronic journals
a. cost assessment
b. price diff.
From these categories, only the costs of work package four (development, design, programming and miscellaneous project costs) are taken into account for the calculation of the yearly costs for EJCL. These costs, totaling NGL 80,666 are depreciated over a four-year period, assuming that every four years, the journals site needs a major overhaul. All the other costs, except of course the costs of work package five (producing the issues), are one-time costs.
- Closesurveys of printed and electronic journals (work packages one an two) are no longer necessary.
- No major changes in the editorial policy (work package three) are expected, incremental changes can be discussed in regular board communications.
- The cost assessment (work package six) was a one time effort.
- Dissemination (work package seven) was a project activity; promotional activities can be carried out by the editorial board and the assistant editor.
The costs of work package five, the actual production of journal issues are replaced by the structural costs of the assistant editor. The Tilburg Faculty of Law has allocated part of its budget to finance a 0.2 fte position. The tasks of the assistant editor are confined to controlling and supporting editorial workflow, copy editing, actual production of the journal and alerting subscribers of newly published issues via the journal's listserver and selected other listservers and Usenet newsgroups. It is expected that 0.2 fte will be sufficient to be able to publish four issues per year with a maximum of four to five extensive articles per issue. The costs of 0.2 fte, including overhead, are NGL 20,000 per year.
In fact, work package five used 379 hours for the production of nearly three issues, which seems at odds with the expectation that 0.2 fte (or 320 hours) should suffice for four issues. This can be explained by the lack of prior expertise in Web-publishing on the part of the assistant editor.
In summary then, the structural yearly costs of publishing EJCL are estimated to be NGL 20,167 (total costs of work package four, depreciated over a four year period) plus NGL 20,000 for the assistant editor, giving a total of NGL 40,167 or $ 20,084 per year.
The cost per issue, if the planned four issues per year are published, will then be $ 5,021. The cost per article published will vary from $ 2,511 to $ 1,004 if respectively two or five articles per issue are published. The last figure is close to Odlyzko's  estimate for mathematical articles, which of course cannot be compared to legal articles.
The ultimate question is, of course, how these figures compare with those of printed journals. The average annual subscription price of all law journals (including free journals and journals published by scholarly societies) in the collection of Tilburg University Library is $175. If the EJCL would have 400 or 600 readers / "subscribers" -- which are not uncommon figures for printed journal subscriptions -- the average annual cost per reader would be, respectively, $50 or $33. Even if we allow for a profit margin of 30 percent on printed journals, which would lead to a net average annual cost of $123, we can conclude that electronic publishing is considerably cheaper. Recovery of these costs is another problem though, with which this article does not deal.
Given the situation of Tilburg University, the main conclusion is that the costs of the EJCL are, in fact, determined by the effort put into the process of developing and maintaining the electronic journal / web site. Notice that the model assumes the site will be redesigned every four years, so the development cost will be repeated.
The costs of shared facilities are minimal because of their relative low usage of existing facilities within Tilburg University. Therefore they are of almost no influence to the outcome.
The outcome might even improve once more electronic journals are developed because of the possible advantages related to the increase of scale, especially for the time spent on the development of electronic journals.
The relative costs per user will decrease substantially once the number of subscribers to the journal grows.
An overall conclusion, with respect to a comparison of the costs of "conventional journals," can only be drawn when an electronic journal has had the chance to establish itself and to outgrow the first "learning phase".
Ellen Finnie Duranceau (ed.), The Economics of Electronic Publishing, Serials Review, 21(1), Spring 1995.
Stevan Harnad and Arnoud de Kemp, Email Debate: Paper Tigers. Academic journals and electronic publishing, ESF Communications, no. 37, October 1997.
Andrew Odlyzko, The Economics of Electronic Journals, First Monday 2(8), 1997, http://www.firstmonday.dk/issues/issue2_8/odlyzko.
Andrew Odlyzko, Tragic loss or good riddance? The impending demise of traditionally scholarly journals, International Journal of Human-Computer Studies, 42, 1995, also available at http://www.research.att.com/~amo.
Carol Tenopir and Donald W. King, Trends in Scientific Scholarly Journal Publishing in the United States, Journal of Scholarly Publishing, April 1997, pp. 135 - 170.
Electronic Review of Comparative Law (preliminary name). Controlling Document, Tilburg University, 1997, available at http://cwis.kub.nl/~dbi/users/roes/erclaw/ercl_ctl.htm.
Hans Roes, Electronic Journal of Comparative Law. Final Report, Tilburg University, 1998, available at http://cwis.kub.nl/~dbi/users/roes/erclaw/ejclfinl.htm.
Marjolein Bot and Johan Burgemeester, Costing Model for Publishing an Electronic Journal, PricewaterhouseCoopers, 1998, available at http://cwis.kub.nl/~dbi/users/roes/erclaw/ejclwp6a.pdf.
Copyright © 1998 Marjolein Bot, Johan Burgemeester, and Hans Roes
Top | Magazine
Search | Author Index | Title Index | Monthly Issues
Previous Story | Next Story
Comments | E-mail the Editor
D-Lib Magazine Access Terms and Conditions