Incubating Open Journals in Economics Malcolm Getz Vanderbilt University Nashville TN 37235-1819 Malcolm.Getz@Vanderbilt.edu Abstract As open journals become intellectual successes, they will come to substitute for subscription-based journals. The entry of two new journals specializing in economic theory, one a commercial subscription journal and the other an open journal, stimulates the question: What does a new journal require in order to achieve intellectual impact? Regressing the ISI Impact Factor for 141 journals on attributes of the journals estimates their relative importance. The intellectual reputations of the editors and of the authors’ prior work along with the number of libraries that hold the title are positively associated with the journal’s Impact Factor. Using the attributes of the new journals in the estimated relationships yields a forecast of the Impact Factor for the new journals. An open journal has the advantage of being immediately as accessible everywhere on the Internet as though held by many libraries. Its forecast impact is comparable to the journals with the highest impact in this specialty. Some open journals are able to attract outstanding editors and authors with the natural advantage of universal access without charge. I appreciate the continuing support of The Cornell University Project on Creating an Open Access Paradigm for Scholarly Publishing and the assistance of Claire Hecht and Ararat Osipian in gathering data. I also appreciate the comments of Paul Gherman, Ken King, and John Siegfried. October 20, 2005 Will open access journals achieve intellectual reputations at least comparable to those of subscription journals? If open journals are intellectual successes, they will come to substitute for some subscription-based journals. Many other changes in the culture of scholarship are likely to follow, including better integration of pre-prints and other archival materials into the universe of scholarship as well as increased international participation in scholarship. The launch of new journals provides an opportunity for weighing the strengths and weaknesses of the two kinds of journals. As new journals of each type appear, there is an opportunity to forecast their likely success and, in a few years, to gauge the accuracy of the forecasts. Economics is a useful arena to consider the incubation of new journals because two new similar journals have entered this arena in 2005-06. One, the International Journal of Economic Theory (IJET), first appeared in March 2005 as a subscription journal published by Blackwell in association with the International Society for Economic Theory. It bears a library rate of $420 per year in the US for print with a premium online service.1 The other, Theoretical Economics (TE), announced in August 2005 with a March 2006 first issue, is an open journal produced by the Society for Economic Theory.2 There is no charge to readers or libraries. Authors pay $75 to submit International Society for Economic Theory (2005). International Journal of Economic Theory. K. Nishimura and M. Yano. http://www.blackwellpublishing.com/journal.asp?ref=1742-7355 . The Premium Online Service authorizes access to electronic archives and in press articles, allows use by walkin clients, and use in electronic reserves and interlibrary loan in addition to the standard online services of electronic access to the published materials via campus network and authenticated IP addresses. 2 Society for Economic Theory (2005). Theoretical Economics. M. J. Osborne. Toronto, Canada. 2005: website for new open journal. http://www.econtheory.org/ 1 essays. There are additional fees to authors if submissions do not conform to standards. TE is produced using the University of Toronto’s digital archiving system with the public domain journal management software, Open Journal System.(Willinsky 2004) TE is available to all readers without charge on the Internet. Print can be provided for extra fees. Both journals have international editorial boards and welcome articles that address a wide range of issues in economic theory. Part of the difference in the new journals may reflect the profit status of their publishers. The average price of a print library subscription to an economics journal with a premium online campus service from a commercial publisher are more than three times higher per page than similar journals from not-for-profits. The average library subscription price of journals published by not-for-profit publishers is $0.31 per page. Journals published by commercial presses other than Blackwell average $0.98 per page. Blackwell, an intermediate case, is a commercial publisher that publishes a number of journals under contract with scholarly societies. The terms of the contracts vary from title to title but may cause some of their prices to be closer to the non-profit publisher rates. For this reason, the prices of Blackwell journals appear as a third category with a mean price of $0.54 per page. Figure 1 shows the frequency distribution of the price per page of economics journal in 2004. (The Journals are those with ISI Impact Factors as explained below.) Successful entry of open-access journals would appear at the zero price point in the not-for-profit set and could provide competition to both commercial and non-profit subscription titles. [See Figure 1.] 2 Which new economics journals will gain the greater intellectual success, the conventional commercial journal or the open-access journal? Because the two new journals target the same, well-defined audience, differences in the intellectual influence of the two journals may be ascribed to pricing and technological differences. Because the two journals are new, neither journal’s stature at launch depends on its own past performance. The analysis here describes the existing journals in economics in order to show the likely position of the new journals. The analysis concludes with a discussion of broader implications for scholarly publishing. Journals in Economics Successful journals have intellectual influence. Well-known authors submit their better works to more influential journals and expect their works to be cited more frequently. A core issue for open access journals is the extent of intellectual influence they are likely to achieve. The ISI Impact Factor is a commonly used measure of intellectual influence. The number of citations to articles published in one year that appear in publications in the two subsequent years is taken as a ratio to the number of articles in the year. The analysis here uses a four-year average impact factor for each journal to better reflect longer-term patterns of influence and abstract from year-to-year variation. The mean impact factor for 154 economics journals is 0.77 with a standard deviation of 0.74. The minimum of the 154 is the Jahrbucher Fur Nationalokonomie Und Statistik at 0.10 and the maximum is the Journal of Economic Literature at 6.0. A core concern for a group of scholars who want to launch their own new journal is how to maximize its intellectual influence. The discussion below considers a variety of 3 features of journals that are likely to be associated with intellectual influence. Few of the features are exogenous. The statistics presented below are descriptive of the patterns in existing journals but cannot be interpreted as reflecting causal relationships. The goal is to anticipate how the new journals will fit among existing journals. EconLit indexes over 400 major journals in economics.3 The ISI Journal Citation Report records an impact factor for 154 economics journals from 2000 to 2003. ISI includes a journal in its report for economics journals when the journal’s impact exceeds a threshold and its focus is primarily economics. This investigation uses information about prices and editors for the 141 economics journals that had Impact Factors reported for four years. The analysis here considers attributes of journals that can be observed at the launch of new journals that relate to a journal’s Impact Factor. Consider the association of observable attributes of existing journals with their Impact Factors, and then assume the same association is likely to apply for the new journals. Use the relationship to compute a forecast of the ISI Impact Factor for new journals and journals that are too young yet to have reported ISI Impact Factors. A principle attribute of strong journals is the intellectual reputations of the editors. Authors are drawn to publish their works in journals whose editors are well known and whose expertise may allow them to suggest improvements in the articles. More editors may mean that an article submitted for publication gets more attention. A larger group of editors may also indicate that the journal attracts more attention. The journals’ websites 3 The AEA website (September 15, 2005) lists 1,100 journal titles as being indexed in whole or in part in EconLit. http://www.econlit.org/journal_list.html Not all are currently published. The list includes titles like the Accounting Review and the Yale Law Journal that, although of interest to economists, would have a primary listing under another discipline. 4 and printed issues provide a list of current editors. The analysis includes a count of editors, including associate editors and members of editorial boards. (Advisory boards are excluded from the analysis.) To measure the scholarly reputation of the editors, we record the mean count of “Cited By” of the five most cited works of each editor as recorded on Google Scholar. The journals averaged 29 editors and the average count of “cited by” is 79 (averaging the average for the 141 journals) as shown in Table 1. A second attribute of strong journals is the intellectual reputations of the authors whose work appears in the journal. Authors are likely to be drawn to submit better works to journals where other influential authors have published their own articles. The analysis includes the average count of “cited by” for the authors whose work appears in a recent issue of each journal. The average number of citations to the authors’ works over all 141 journals is 35. A third attribute is the presence of the journal in the library collections of the world. Libraries play an important role in providing access to scholarship for faculty, students, and others. A journal that is held by more libraries is more likely to have its articles be more widely cited. The OCLC WorldCAT service provides a count of libraries that have posted their ownership of the title (not necessarily current receipt) in the OCLC database. Approximately 40 percent of libraries post titles held in OCLC.4 The OCLC count is likely to be correlated across journals with the total number of libraries that hold the journal. The average WorldCAT count of libraries that hold the journal for the 141 About 41 percent of libraries record their serials holdings in OCLC’s database. This estimate derives from comparing the 1,841 libraries reporting holding the American Economic Review to the 4,482 subscriptions reported by the American Economic Association (AEA). Some of the AEA’s subscriptions are to banks and research institutes that may be less likely to participate in OCLC than academic and public libraries. 4 5 journals is 416, a number consistent with about 1,000 libraries holding the average journal. OCLC member libraries include academic, public, and business libraries as well as research institutes, in the US and around the world.5 The price of the journal to libraries is another way of measuring likely access through libraries. Because the OCLC holdings measure is imperfect, the price of the journals to libraries may correlate with the number of libraries who hold the title currently. The library price used in the analysis is the rate that includes premium campus online access when it is available. A few of the journals do not offer an online service and are omitted from the analysis. The average library price is $701 including online access. The correlation between the WorldCAT count and the library price is -0.18, an indication that price is measuring something different than the count of libraries. The library price correlates with personal subscription rates. Twelve of the 141 journals have a submission fee for at least some authors. Higher submission fees might discourage authors from submitting their works and so tend to lower the intellectual influence of the journal. The analysis uses a binary variable to flag the presence of a submission fee rather than including the magnitude of the fee itself because the variety of structures of fees for authors is not easily captured in a continuous variable. Some author fees are waived for members; some are waived for authors who review for the journal. Author fees may vary depending on the format of the submission, as with TE, and with page charges. A recent survey of publishers across disciplines indicates that author charges are commonly used by both subscription and open journals.(Morris 2005) 5 More than 53,000 libraries of all kinds in 96 countries and territories participate in OCLC’s services. http://www.oclc.org/about/default.htm 6 Journals published by not-for-profit publishers may have advantages in creating influence. Forty of the journals are not-for-profit. Not-for-profits might be more effective in creating intellectual influence, for example, in attracting donated services from editors and reviewers. Alternatively, not-for-profits might be less efficient if consensus among a large, volunteer board is needed to make major decisions concerning the journal in contrast to the possibly more decisive action of a for-profit entity. The age of a journal may also affect its influence. Because reputations take time to build and tend to persist, older journals may have an advantage. Thirty-one of the journals predate 1960. At the same time, a new journal might catch new trends or more easily achieve a sharper focus that authors and readers find appealing. Five journals postdate 1995. Table 1 shows the mean year of launch for the journals was 1969 (median is 1977). The analysis includes a binary for launch before 1960 and another for launch after 1995. Preliminary analysis showed that the Journal of Economic Literature is an outlier. It publishes only commissioned review articles and gets its own binary in the analysis. The number of pages published per year is included. A larger journal may draw more attention to the articles it publishes. The journals average 956 pages per year. Thirty-three of the journals are general interest journals, that is, they have no specialty.6 The other 108 journals group into 19 specialty groups. Eight journals are in economic theory, the specialty of the two new journals. Three journals are in econometrics and four are in macroeconomics, two specialties often viewed as in the core Dhuey, E. and T. C. Bergstrom (2005). New Data on Journal Prices in Economics. Santa Barbara, CA. http://www.econ.ucsb.edu/%7Etedb/Journals/pricing.html 6 7 of the discipline. Six are in labor economics, a specialty that preliminary analysis showed had differential impact. On average, articles in a recent issue of each journal had already been cited about seven times at the time of publication.7 Essays circulate as working papers, are presented at workshops and conferences, and are posted on the Internet, allowing citation before publication. Formal publication ratifies the work but is often not the starting point for its influence. This variable doesn’t appear in the analysis but receives more comment below. The character of editing provided by the journal would be an appropriate attribute of a journal to include in the analysis but no measure is available. Some journals have well-paid editors, some use only volunteers. Lead editors manage the process of deciding what to accept for publication. A significant amount of editorial services occurs after the decision to publish. A recent survey of groups of publishers reported that as many as 60 percent of some groups of journals edit content.8 Among 34 journals associated with the Association of American Medical Colleges, about 25 percent edit content. Nearly all journals, for example 96 percent of the 34 journals in the Association of American Medical Colleges group, 7 At the time of the survey in the summer, 2005, we took the most recent issue of each journal and noted the number of “cited by” items at Google Scholar for each article. The count was typically within three months of publication, too soon for authors of the citing works to have read the article in publication, referenced it in a new work, and have had the work be published. Note, however, that some journals now post articles online “in press” some months in advance of the official publication date. 8 The Association of Learned Professional and Scholarly Publishers (ALPSP) in the United Kingdom reports that 62 percent of the 128 journals published by its members edit content. Among the 85 journals published by the HighWire Press at Stanford 39 percent edit content. Among 248 journals surveyed from the the Directory of Open Access Journals, 34 percent edit content.Morris, S. (2005). The facts about open access. Clapham, Worthing, West Sussex, UK, Association of Learned and Professional Society Publishers. 2005. http://www.alpsp.org/publications/FAOAcomplete.pdf Table 36, page 48. 8 provide copy-editing to improve the grammar and style of their articles; a few do not. Including in the analysis information about the extent of editing provided by the journal would allow estimating how editorial services relate to intellectual impact. Presumably, more extensive editing increases readership and influence. A closer look at the journals in economic theory shows where the new journals in this subfield will fit. Table 2 reports details for a prominent general journal for comparison, nine existing journals in economic theory, and the two new journals. [See Table 2.] The description of the ten established journals raises the question as to the demand for new journals in this arena. The opening editorial in IJET points to a “serious shortage of both journal pages and journal selection that exists in the field of economic theory.”(Nishimura and Yano 2005) Yet, the two young journals in this field are more than five years old, are held by relatively few libraries, and have so far not achieved sufficient intellectual effect as to be indexed by ISI. The number of pages in the youngest journal, International Game Theory Review, is below that of all the others. Several of the existing journals could double their page count and still be within the range of pagecounts of the titles in the group. The price of IJET at launch is well above that of the two young journals. In contrast, TE justifies its launch with the belief that “Open Access is the right way to disseminate scientific information.”(Society for Economic Theory 2005) It makes no claim that important essays remain unpublished, but rather says that open access will give authors “maximum possible exposure for their work.” TE then intends to substitute open access publication for conventional subscription-based publication. 9 The central question for success in each case is likely to be whether the journals will attract quality work from economic theorists. If the founders of IJET are correct, the hoped-for growing influence of their new journal will not reduce the influence of related journals, presumably because good quality work is going unpublished. If the founders of TE are correct, their journal will attract better quality articles away from existing journals. TE’s rising influence will come at the expense of lower influence of other journals. The established journals then define the playing field. Will the new entrants succeed by producing journals of greater influence than some existing journals? Will the new open-access journal be more influential than the new subscription journal from a commercial press? Patterns in Impact Factors The analysis here considers how the Impact Factor of journals relates to their other attributes. A regression of the Impact Factor on the number of editors, the average rate of citation to the editors’ five most cited works, and the number of libraries in OCLC reporting that they hold the journal, and the other variables mentioned above appears in table 3. [See Table 3.] Because causal relationships go in both directions, these statistics are simply descriptive rather than providing a test of causality. Libraries are more likely to buy journals with high impact factors just as high impact factors are likely to be generated by journals that are more widely read when found in more libraries. Editors with stronger reputations are more likely to be drawn to work for journals with higher Impact Factors just as more capable editors are likely to produce a journal with more impact. Even 10 though we cannot disentangle causality, the regression describes a pattern among journals. The regression reported in table 3 shows a strong relationship among the variables in line with expectations. The Journal of Economic Literature is treated separately as an outlier with its own binary that shows a large positive effect. The intellectual reputations of the editors as indicated by the average “cited by” counts in Scholar have a strong association with the ISI Impact Factor as does the “cited by” count for previous work by the authors in a recent issue. The WordCAT count of libraries holding the title also has an association with impact. Editors with stronger intellectual reputations are associated with journals with higher impacts. Authors with stronger intellectual reputations publish in journals with higher impact factors. When more libraries hold a title, the journal’s impact is higher. Price and fees, number of pages, the age and profit status of the journals aren’t associated with the impact factor, other things equal. Given the other features of the journals included in the analysis, journals with higher prices are associated with neither higher nor lower impacts. This fact bodes well for open-access journals. If higher subscription revenues were associated with more intensive editorial services that yielded more impact, then open access journals would be climbing uphill. With impact not associated with price, the open access journal may hope to have significant intellectual impact with modest revenue streams, provided it has a strong editorial group. The fact that journals with more pages are not associated with higher intellectual impact, other things equal, also bodes well for open-access journals. A journal can achieve significant intellectual impact without publishing a large volume of pages. 11 Perhaps quality and focus matter as much or more than scale. A new entrant may more readily pursue quality and focus than scale. Similarly, the lack of influence of age suggests that new journals face no significant reputational disadvantage for being young. That author fees are not associated with lower impact, other things equal, also suggests that author fees may be a possible source of revenue. So few existing journals have such fees and the nature of the fees vary enough that conclusions here may be unwarranted. That the profit status of the journal is unrelated to impact, other things equal, suggests that the for-profit and non-for-profit publishers work within common constraints. Not-for-profit journals have an important place. The five binaries for fields treat all other fields as the omitted set. General interest journals show no different impact than the omitted groups nor do the econometrics journals. Interestingly, journals in economic theory show a statistically significant lower impact factor, other things equal, as do the macroeconomic journals. The journals in labor economics show a positive bump in impact. Each field may have its own conventions in terms of citing other material. If each article cites fewer other works, citation rates in a subfield will be lower than average. Forecasting Impact Factors The estimated relationship shown in table 3 allows a forecast of ISI Impact Factor for journals for which the Impact Factor isn’t reported, including new journals. The launch of IJET reveals information about its editorial board, authors, prices, and page counts. It’s 24 editors show an average “cited by” count of 19 and its authors show a cited by average of 30. The forecast of its ISI impact factor using the estimated regression relationship is 0.22. This value is likely to grow if library holdings increase. 12 [See table 3.] In contrast, Theoretical Economics has three advantages at launch. It has a substantial number of editors, 42. The average “cited by” rate for its editors is 67, more than triple that for the editors of the new commercial journal. The authors in the inaugural issue have a cited by rate of 90 that is also triple that of the new commercial journal. Open access provides instant availability everywhere, probably better than being held by subscription in 500 libraries. (Open-access means essentially universal availability; the 500-library assumption is an assumption about the maximum number of institutions with interest in economic theory at a zero price.) The given attributes of the TE journal at launch yield a forecast for its ISI of 0.71. This forecast rate is similar to the 0.75 Impact Factors of the Journal of Economic Theory and Games and Economic Behavior, the leading journals in this specialty. The submission fee for TE is similar to that of other good quality society publications and the regression does not signal that a submission fee at going rates is associated with an adverse effect on expected Impact Factor. A recent survey of publishers reports that requiring fees from authors encounters little resistance.(Morris 2005) The same method provides forecasts for Impact Factors for the two young journals in the same specialty that did not have published scores. The forecast for the Review of Economic Design (Springer is the publisher) is 0.67 and the forecast for the International Game Theory Review (World Scientific is the publisher) is 0.23. At launch, then, the open journal appears to have better prospects than the new subscription journal. 13 Worldwide Impact Three-fourths of the American Economics Association’s revenue from subscriptions to its index database, EconLit, is from countries outside the US. Scholarship is increasingly international. Open journals have a significant advantage in being instantly available everywhere on the Internet without fees. Scholars all over the world have ready access to open journals. As a consequence, citation rates are likely to grow faster among open journals than among subscription journals. Open journals have a significant advantage in attracting readers, authors, and editors everywhere on the Internet simultaneously. This effect may cause the estimate here to understate the likely impact of the open journal. Google Scholar and ISI Impact The exercise presented here uses information from Google Scholar and ISI as alternate but comparable measures of scholarly impact. The two sources of data about publishing differ in their scope and accuracy. To judge how they relate, consider using Google Scholar to compute an impact factor that is comparable to the ISI Impact Factor. As explained above, ISI uses its database of citations to compute the number of citations to all the essays in an annual run of a journal that appear in its database in the two years following the year of publication as a ratio to the number of articles in the year. Google Scholar provides advanced search tools that allow one to compute the number of citations in its citation database in the two years after publication and compute a score that is the citations per article. Searching for the number of citations in Google required searching each article individually, a time consuming task. (Perhaps Google will automate it.) The analysis here 14 makes the comparison based on the comparable scores for 25 randomly selected journals in economics for which ISI reported Impact Factors in 2003. The mean ISI Impact Factor for the 25 was 0.836 and the mean for all 168 economics journals with scores in that year was 0.765. The mean for the Google analog for the 25-journal sample was 3.963. (The Google scores used here are for articles published in 2002.) On average Google gives nearly a five times larger count. Its mechanical search of the Internet casts a wider net and includes working papers, government documents, books, and other works not included in the ISI database. Google’s search may also be sloppier with the inclusion of inaccurate and duplicate references, an issue not explored here. A regression of the Scholar-based factor on the ISI Impact Factor explores the comparability of the two measures. The results, with t-ratios in parentheses, indicate that, on the margin, there are 7.549 citations per article in the Google factor for each citation in the ISI factor. The r-square is 0.667. In broad strokes, the Google Scholar impact factor and the ISI Impact Factor are similar but not identical. GoogleFactor = -0.966 + 7.549 (-0.873) (6.78) ISI Factor Two journals in the sample of 25 have substantial influence on the estimate. Health Economics has an ISI factor of 1.713 that is high relative to the Google factor of 3.280. Econometrica’s ISI Factor of 2.737 is low relative to its Google factor of 22.244. Perhaps Google’s universe captures more working papers in some domains than others and perhaps scholars working in some domains show somewhat different behavior in posting essays as working papers and in the number of citations per essay. Omitting these 15 two outliers, however, changes the regression result very little, yielding a coefficient of 8.93 with a t-ratio of 6.00 (r-square = 0.61). More detailed investigation might explore the extent to which the differences are the result of scope and accuracy. Open journals have the advantage of instant full-text indexing via Google Scholar. Pre-print Effect A possible additional advantage for an open journal is the more frequent and reliable access to preprints in open archives. Authors who aspire to publish their works in open journals have less reason to refrain from posting works in progress as preprints. Perhaps essays that are published in open journals will be more likely to have been cited as preprints before publication in a journal. At the time of publication, of the 141 journals, 28 showed no citations to essays as they were published. The average rate over all the journals was seven citations to the essays at publication. The highest average was the American Economic Review (AER) with an average of 178 citations to the articles as they were published. The AER had a reputation of having long publication lags that would tend to allow citations to working papers to accumulate before publication. Clearly, the culture of sharing working papers is well established in economics and provides a natural advantage for open access journals that have fewer reasons to suppress access to working papers or the articles themselves in any form. The rate of citation to the essays in TE’s first issue won’t be known until their publication date in March 2006. ISI reports as an Immediacy Factor, the number of citations to works in the same year as the publication appears. The mean for Immediacy is 0.163. A regression of the Scholar count of “cited by” for an issue at publication on the Immediacy Factor has an r- 16 square of 0.039 with coefficient on Immediacy of 28.75 (t = 2.03). Immediacy measures something different than the Google Scholar measure. Google indexes working papers and this may explain its much higher level of magnitude and low correlation with Immediacy. Regressing Impact Factor in a year on Immediacy and the Google “cited by” at publication yields an adjusted r-square of 0.463. The Immediacy Factor has a coefficient of 3.60 (t= 8.79), an indication that Immediacy tracks Impact. The Google cited-by-at-publication has a coefficient of 0.005 (t = 1.74), an indication that the Google count has only a modest effect in forecasting Impact. The investigation reported here used only one issue of each journal in computing the Google “cited by” at publication. Measuring the count requires doing it for each issue as it appears because Scholar continuously updates. Implications The analysis above demonstrates that new open journals are capable of attracting a strong group of editors and are likely to generate high impacts within a relatively short period of time. The new open journal in economic theory, TE, appears to launch with a much higher probability of success than its subscription-based new competitor, IJET. Indeed, its prospects are high relative to the established journals in this subdiscipline. The forecast here of success for the open journal in economic theory is consistent with the success shown for open journals in other fields. A citation study from Thompson Scientific indicates that open journals are growing in number and many are enjoying significant intellectual impact in medicine, life sciences, physics, engineering, and mathematics.(McVeigh 2004) Although the details may differ, journals with strong editorial boards that attract frequently cited authors and offer universal availability 17 through open access are likely to achieve significant intellectual impact quickly. Although the analysis here looks only at journals in economics, the factors shown to be important are likely to hold sway for journals in many disciplines. Suppose, as seems likely, that the claim made by the editors of IJET noted above is incorrect; that is, assume that most worthy essays in economic theory are already readily published. Then, the success of the new open journal will come at the expense of the influence of the existing journals. Authors who place their most promising work with the open journal will submit fewer and less promising works to the other journals. Although the journals in this field aren’t perfect substitutes intent only on maximizing impact (they each may emphasize particular issues and reflect the tastes of editors and authors), nevertheless, many authors may decide, on the margin, to prefer publication in journals with more impact. Journals of lesser reputation in the economic theory field may lose impact as Theoretical Economics attracts more influential authors. To test this hypothesis, it will be useful to track how the impact factors for all the journals in the field adjust over the years immediately ahead as the new entrants gain reputation. At this point, we offer some speculation. [See Figure 2.] Figure 2 shows where the new entrants and young journals fall relative to the established journals in terms of price per page (library rate with print and premium online service) and ISI Impact Factor (forecast for those not published). The overall relationship shows journals with higher prices per page having lower impacts. (Note that unlike the regression reported above, Figure 2 does not control for other features of the journal.) Games and Economic Behavior—GEB—has a relatively low price with a high impact. 18 The Journal of Mathematical Economics—JME—has a high price and relatively low impact. Elsevier publishes both. One of the young journals, International Game Theory Review—IGTR from World Scientific, is an outlier with a forecast for Impact Factor that is lower than its price alone would suggest. The Journal of Economic Theory—JET, also from Elsevier, has a somewhat higher impact factor relative to its price than the general pattern. The two new journals enter at each end of the range. The new commercial journal, IJET, has a high price per page and a low forecast for Impact. The new open access journal, TE in the figure, has a zero price and a forecast Impact Factor that is among the top in this specialty at launch. If the assumptions here are correct, an open access journal with a strong editorial group can launch with an expectation of high impact with low price. All open-access journals should have the natural advantage of being as available as a journal that is in every library, perhaps even more so because it is available on every desktop. Assuming that the number of journals doesn’t influence the rate of creation of new essays in this specialty, the Impact Factors of existing journals with high prices and limited library access should decline. The relationship shown in the figure will become steeper. It should be particularly difficult for a new entrant like IJET with a high price per page to gain intellectual influence. The Journal of Economic Theory (JET in the figure) has a high price and strong reputation. If the open access Theoretical Economics meets and sustains the forecast given here, the Impact Factor for JET will decline. Successful open-access journals are likely to bring a more pronounced tilt to the price—Impact Factor relationship. 19 Ultimately, some of the weaker journals, mostly those with high prices and low intellectual impacts, will fold. The success of a significant open access journal in economic theory might encourage entry of other open access journals in economics. Indeed, there may be room for additional entry by other open journals in the economic theory field, focused on narrower subspecialties. Open journals should be attractive to editors because readership is not tied to subscription sale. Content will be readily accessible with full-text searching of Google Scholar and other indexing services should quickly begin to index journals of substantial intellectual impact. Authors may promote their work as preprints without fear of loss of opportunity to publish in highly regarded journals. The significant rate of citations before publication noted above indicates the importance of preprints in building reputations. The unmeasured dimension of the journals is the effort put to providing editorial services as noted above. Open access journals are more likely to rely on volunteer managing editors and less likely to provide content and copy editing because they generally operate with lower revenues. Advertising, institutional support, and author fees may support, particularly, editorial services applied to articles after the decision to publish them. Universities Successful entry by more journals will depend on universities sustaining the necessary infrastructure and making it available to well-qualified scholarly groups who will introduce and sustain open journals. University libraries are possible agents. There is some fixed cost in setting up the infrastructure for archiving digital files and in 20 supporting the tools for managing the flow of materials through the editorial process. The cost of supporting additional journals once the infrastructure is in place is modest. Research universities that make significant expenditure on digital infrastructure to sustain their research missions are likely to be willing and able to support open access journals. Managing Editors The critical actors are the leading scholars who edit journals. Will they find it attractive to serve as lead editors for open access journals? The launch of Theoretical Economics demonstrates the possibility. However, commercial and established society publishers recruit and pay lead editors for their work. The open-access journal depends on editors being supported by means other than revenue from journal subscriptions. Theoretical Economics has volunteer editors, which might more properly be viewed as editors supported directly by their home universities rather by the journal’s revenue. The central challenge is in recruiting and sustaining lead editors, one to three people who will manage the editorial process and make final editorial decisions. Suppose, say, four dozen leading scholars with shared intellectual interests decided that they would be better served by an open-access journal than conventional subscription journals. They value the immediate, Internet-wide ease of access, participation in Google Scholar, and relatively seamless integration of working papers with published work that an openaccess journal provides. They associate themselves as a professional society and anoint a core group to serve as a board of directors to launch an open journal to serve their field. All of them have agreed to serve as volunteers on the editorial board with a commitment to review submissions and submit their own work and to encourage their students and colleagues to submit their work for publication. The board of directors must find the 21 scholar among their number who will be the lead editor with perhaps one or two coeditors. The lead editor will need to be on a campus that will provide the infrastructure for the new open journal without charge to the new scholarly society with the agreement that the journal will be open and that the university will get prominent mention for its services. The lead editor must also have the support of her or his department and dean for taking on the role. The university’s infrastructure and support for a potential editor probably come as a package. When an editor’s tour comes to an end, the board will seek a replacement editor, presumably with backing from his or her home university. The launch of Theoretical Economics demonstrates that this process can lead to a successful launch. An alternative to direct university support is submission fees from authors as used by the Public Library of Science (PLoS) for journals in medicine and biology. The PLoS journals charge $1,500 submission fees and devote resources to copy-editing that yield higher production values. Higher production values allow work to reach a wider audience. In contrast, Theoretical Economics, like most academic journals, will depend on authors to prepare their work in light of comments by reviewers and editors. Theoretical Economics’ submission fee will support reviews by volunteers. Author submission fees and page charges are likely to continue to be more important for journals in disciplines where most research is funded from outside universities. Author fees are typically a small fraction of the total size of grants. In disciplines where funded research is less common, author fees are likely to be less successful. Basic copying editing and manuscript preparing may cost $19 per page.9 At 9 An anonymous publisher (private communication) pays $19 per page for manuscript preparation for one of its journals. The service includes: page makeup tabular, math, captions, references, rules, boxes (if included in supporting data), rapid proofs, pick up of 22 this rate, a typical essay of 20 pages would require $380 in copy-editing and manuscript preparation. (With more equations and tables, the cost will be over $600 per article.) A journal with a strong intellectual reputation might invite authors to put their accepted manuscripts in the necessary format or pay appropriate fees of this order of magnitude to reach standards. Tying specific author fees to specific services would allow authors to find an alternative or pay the fee. The submission fees might support the review process while page charges pay for post-acceptance editing. Some journals pay reviewers and some waive submission fees for authors who have reviewed other essays for the journal.10 Whether higher author fees can support more sophisticated editing of content is an open question. The Public Library of Science appears to offer more than basic copy-editing with a publication fee of $1,500 per article. An analysis of the revenue and cost structure of academic journals indicates that current library expenditures on journal subscriptions and associated support costs within libraries are well above the cost of editing and producing the journals.(Getz 2005) A shift to open-access journals could reduce the total flow of dollars from university budgets that support journals without reducing expenditures on editorial services. This is possible because of the significant level of profit in commercial publishing, the lower shipping and transactions costs in making journals available at zero price, and the reduced costs to standing pages, ASCII page processing, PDF page processing for archives, and archiving. The rate for a more complex journal is $31.50 per page. 10 A colleague mentioned receiving a fee of $500 for preparing a review of an article for a journal in economics. A submission fee of at least $1,000 would be necessary to support two reviews per article at these rates. If associate editors add a third review, a submission fee of $1,500 would be needed. Presumably such reviewer fees encourage speedy review. Submission fees also play the role of discouraging casual submissions. A journal that receives many more submissions than it can manage with volunteer reviewers may impose a submission fee to limit submissions to those authors view as most promising. 23 libraries of avoiding payment systems.11 Open journals do not require the level of revenue that commercial journals now receive. Going further, the shift to open journals with university-based infrastructure may lower the explicit university expenditures that now sustain the editorial costs through subscriptions, at least for many academic journals that do not provide content editing beyond what reviewers provide. The universities may find it more attractive to bid for and sustain the scholars who are editors directly rather than using publishers as intermediaries. Scholarly Societies Open journals pose a challenge to traditional scholarly societies. The established societies depend on subscription revenue from libraries for a significant part of their revenue. They also earn revenue from individual memberships that reflect primarily journal subscriptions to individuals. Some receive advertising revenue that might be reduced in open access, digital only journals. If open journals replaced subscription journals, the scholarly society might continue to offer meetings and other member services but would probably be smaller organizations. Scholarly societies do not have a significant source of revenue other than from subscriptions and advertising that would allow them to pay for the infrastructure and overheads that are needed to sustain open 11 Individual universities where faculty members publish at high rates could see a higher expenditure on author fees than for journal subscriptions. The author fees, however, should be considered as net of support of the fees paid by grants. On a net basis, even universities with the most prolific faculty are likely to see a lower commit of core university funds to the support of journals with open access and author fees than with subscriptions. Granting agencies might make clear their support for open publishing in their program announcements and funding policies. 24 journals. A fortiori, a society cannot convert from subscription to open journals and generate surpluses to support other society activities. The two new journals considered above were launched by new societies that appear to have formed for the purpose of launching each journal. These societies appear to have no other function. Neither society will subsidize its journal from other society activities nor will the journals generate surpluses that the society might use to support other functions. These societies will likely rise and fall with their respective journals. Whether a traditional society might sustain premier journals with subscription and author fee revenue and at the same time support other open journals using volunteer editors and university infrastructure is an open question. There is a limit to the amount of editorial services volunteer editors will provide. Premier journals may require subscription revenue to support more on editorial services and thereby sustain a high level of readership. The lack of information about the editorial services provided by the journals in the analysis above precludes statistical inference here about the role of editorial services in enhancing readership and influence. It is possible that a traditional society will continue to use subscription revenue to support a high level of editorial services with its premium journals that co-exist with open journals that do no postacceptance editing of content. The recent survey of journal publishers suggests that the value of copy-editing is in decline among subscription journals.(Morris 2005) Not-for-profit publishers have an advantage in being more plausible partners for universities who support open journals edited by their faculty. Historically, universities supported conventional print, subscription-based journals through their university presses. Many journals were closely associated with their university hosts and some still 25 are. For example, The Journal of Political Economy is published by the University of Chicago Press and is identified with the University. The universities gained reputation in advancing scholarship through their journals. Hosting a journal in a field may have been of value in recruiting faculty who published in the journal or who were editors. The success of open journals may depend, in part, on research universities’ willingness to provide the infrastructure for open journals and to support the editors who produce distinguished journals. In some cases professional societies sustain partnerships with universities who host their journals. For example, the Virginia Commonwealth University currently helps support the Southern Economic Journal. The professional society might, in effect, franchise its name with a quality control mechanism to affiliated universities who provide the infrastructure and support the core editorial services for open journals. The professional society can move a successful journal from one campus to another should a university host lose interest or quality slip. An important choice falls to authors who weigh the advantage of likely greater intellectual impact of open access but with editorial services limited to volunteers with direct university support plus editorial services that author fees will support against a probable lower level of intellectual impact with subscription-based journals but higher level of editorial services. Editorial services, of course, are of different kinds. At one level, subscriptions allow a publisher to pay higher salaries and so bid for more talent in lead editors. At another level, publishers pay for rewriting, copy-editing, manuscript preparation, and design that create more readable essays. Even in the subscription world some journals performed reasonably well with volunteer editors, with editors contributing 26 their services with the encouragement of their universities in order to promote scholarship in a community they valued. Subscription revenue paid for printing, mailing, and some clerical support. Open access would appear to dominate this type of journal. At the same time, some journals engage well-paid editors and provide a significant level of manuscript development to produce journals that are more widely read as a consequence. Subscription revenue may remain an effective way of supporting such journals. There is then likely to be co-existence with a shift over time toward open journals as they realize their potential. The publication of scholarly journals is in flux caused by digital publishing and the Internet. In many disciplines, journal articles that are not readily available on the Internet are less likely to be read and cited.(Kurtz 2003; Brody 2004; McVeigh 2004) As open journals succeed, perhaps even campus site licenses will be a deterrent to use. Some publishers are experimenting with delayed open access, that is, allowing their journals to be freely available on the Internet after six months or a year of publication. Others allow an author to pay a fee, for example $3,000 per article at Springer, to have his or her essay made freely and immediately available on the Internet.12 The survey of publishers indicates participation by up to 25 percent of authors at some journals.(Morris 2005) These methods of providing partially open access seem likely to be intermediate steps in a period of rapid evolution. Open journal have natural advantages in instant, worldwide access with low distribution costs. Universities and scholarly societies who can find ways to underwrite the infrastructure and sustain quality editing are likely to enjoy increased intellectual 12 Springer Open Choice http://www.springeronline.com/sgw/cda/frontpage/0,11855,440359-0-0-0,00.html 27 influence. There will most likely be a range of journal types with revenue sources and editorial services that vary substantially across disciplines and within disciplines as well. Each discipline has journals distributed over a range of intellectual influence. The entry of Theoretical Economics along with some quite influential open journals in other disciplines demonstrate that open journals will be found across most of the range of influence. Authors and potential editors appear to be ready to take advantage of the open opportunity where research universities provide the infrastructure and support their faculty who are lead editors. Where philanthropy, government support, advertising, and industry support are available, open access journals are likely to advance rapidly. Open journals are moving well beyond the fringes of scholarly communication and into the mainstream. References Brody, T., et al. (2004). The Effect of Open Access on Citation Impact. http://www.ecs.soton.ac.uk/~harnad/Temp/OA-TAadvantage.pdf Dhuey, E. and T. C. Bergstrom (2005). New Data on Journal Prices in Economics. Santa Barbara, CA. http://www.econ.ucsb.edu/%7Etedb/Journals/pricing.html Getz, M. (2005). "Open-Access Scholarly Publishing in Economic Perspective." Journal of Library Administration 42(1): 1-39. http://dspace.library.cornell.edu/bitstream/1813/177/2/Getz_11Jun04.pdf International Society for Economic Theory (2005). International Journal of Economic Theory. K. Nishimura and M. Yano. http://www.blackwellpublishing.com/journal.asp?ref=1742-7355 Kurtz, M. J. (2003). Restrictive Access Policies Cut Readership of Electronic Research Journal Articles by a Factor of Two. OpCit The Open Citation Project. Highfield, Southhampton, UK. http://opcit.eprints.org/feb19oa/kurtz.pdf McVeigh, M. E. (2004). Open Access Journals in the ISI Citations Databases: Analysis of Impact Factors and Citation Factors. Philadelphia, PA, Thomson Scientific. http://www.thomsonisi.com/media/presentrep/essayspdf/openaccesscitations2 .pdf 28 Morris, S. (2005). The facts about open access. Clapham, Worthing, West Sussex, UK, Association of Learned and Professional Society Publishers. 2005. http://www.alpsp.org/publications/FAOAcomplete.pdf Nishimura, K. and M. Yano (2005). ""Editorial"." International Journal of Economic Theory I(1): 1. http://www.blackwellpublishing.com/journal.asp?ref=1742-7355 Society for Economic Theory (2005). Theoretical Economics. M. J. Osborne. Toronto, Canada. 2005: website for new open journal. http://www.econtheory.org/ Willinsky, J. (2004). Public Knowledge Project Open Journal System. Vancouver, BC, University of British Columbia. 2004: open source software for journal management. http://www.pkp.ubc.ca/ 29 Table 1 Average Values for 141 Economics Journals Variable ISI Impact Factor, 4-year average Mean StdDev 0.80 0.74 Number of Editors 29.35 15.16 Citations to Editors’ Works 78.82 73.78 Citations to Authors’ Works 34.84 44.47 WorldCAT Libraries, count 415.51 343.05 Library Price w/online, dollars 700.92 710.57 Submission fee=1 0.09 0.28 Not-for-profit=1 0.28 0.45 Before 1960=1 0.22 0.42 Post 1995=1 0.04 0.19 Journal of Economic Literature =1 0.01 0.08 956.40 540.04 1969.24 25.62 Cites at Publication 6.93 19.53 General Interest = 1 0.23 0.42 Theory = 1 0.06 0.23 Econometrics = 1 0.02 0.14 Macroeconomics = 1 0.03 0.17 Labor = 1 0.04 0.20 Pages in Year (100s) Year Began Source: Author’s calculations from ISI Journal Citation Reports, OCLC’s WorldCAT, and journal websites. 30 Table 2 Journals in Economic Theory Launch Pages Journal Library Publisher year 2004 price price/page WorldCAT Econometrica Society 1932 1,935 $550 $0.29 789 Journal of Economic Theory Elsevier 1969 2,041 $2,498 $1.07 503 Springer 1971 600 $657 $1.10 198 Elsevier 1974 962 $1,571 $1.63 245 Elsevier 1979 2,552 $1,470 $0.82 211 Elsevier 1980 1,876 $1,547 $0.97 232 Behavior Elsevier 1988 1,738 $700 $0.40 169 Economic Theory Springer 1992 1,500 $1,219 $0.81 107 Review of Economic Design Springer 1996 477 $312 $0.65 83 Scientific 1999 395 $267 $0.68 28 Blackwell 2005 310 $402 $1.30 4 Society 2006 560 $0 $0.00 500 International Journal of Game Theory Journal of Mathematical Economics Journal of Economic Dynamics & Control Journal of Economic Behavior & Organization Games and Economic International Game Theory Review World International Journal of Economic Theory Theoretical Economics 31 Source: OCLC World Cat, Journal Websites, author’s calculation. Library Price is the price of print with a premium online license, if offered. Econometrica price is 2005 from website. “WorldCAT” is the number of libraries with the journal title in its catalog as listed on OCLC’s WorldCAT in May 2005. Multiply by 2.5 to estimate number of libraries that hold the title. Econometrica is a general journal, shown for comparison. The page counts for the International Journal of Economic Theory and Theoretical Economics are extrapolations from the first issues. The WordCAT count for Theoretical Economics is a rough guess as to the consequence of unlimited Internet access. 32 Table 3 Regression of ISI Impact Factor Variable Coefficient t-ratio Constant 0.064 0.557 Number of Editors 0.002 0.747 Citations to Editors’ Works 0.004 *** 6.070 Citations to Authors’ Works 0.003 ** 2.490 WorldCAT Libraries, count 0.001 *** 3.060 Library Price w/online, dollars -0.009 -1.020 Submission fee = 1 -0.183 -1.270 Not-for-profit = 1 -0.060 -0.562 Before 1960 = 1 0.043 0.362 Post 1995 = 1 0.305 1.360 Journal of Economic Literature =1 3.606 Pages in Year (100s) 0.014 1.140 General Interest = 1 -0.065 -0.628 Theory = 1 -0.368 Econometrics = 1 -0.251 Macroeconomics = 1 -0.506 * -1.970 0.364 * 1.940 Labor = 1 *** ** 6.620 -2.080 -0.874 n = 141. Adjusted r-square = 0.656. s = 0.4357. ANOVA F(16,124) = 17.7. *** Statistically significant at the 0.01 level; ** at the 0.05 level, * at the 0.10 level, two-tailed tests. 33 Figure 1 Economics Journals: Price per Page Frequency of Journala 35 30 25 20 15 10 5 0 $0.00 $0.50 $1.00 $1.50 $2.00 $2.50 $3.00 Price per Page 2004 not profit Blackwell Other Commercial Source: Author’s calculation using data from the publications. Figure 2 ISI Impact Factor ISI Impact and Library Price 0.8 TE 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 $0.00 GEB JET JME IGTR $0.50 IJET $1.00 $1.50 $2.00 Library Price per Page New Journals Young Journals Established Journals Source: Author’s calculation using data from the publications. 34