The Bundling of Academic Journals Aaron S

Berkeley Law

From the SelectedWorks of Aaron Edlin

June, 2005

The Bundling of Academic Journals Aaron S. Edlin Daniel L. Rubinfeld

Available at: https://works.bepress.com/aaron_edlin/40/ COMPETITION POLICY FOR JOURNALS†

The Bundling of Academic Journals

By AARON S. EDLIN AND DANIEL L. RUBINFELD*

Academic journal publishing has evolved rap- strategic barrier has emerged. Major publishers idly in the past two decades. Prices, ownership have been offering libraries packages of jour- concentration, the number of journals, and the nals that are bundled across journals and across means of distribution have all changed dramati- print and electronic versions. The exact terms cally. Substantial price increases have been the have varied from publisher to publisher, but a norm. Average prices have risen severalfold over contract (sometimes called a “Big Deal” by this period, with prices climbing the most at for- librarians) typically involves a library entering profit journals, where these prices are now as into a long-term arrangement to get access to a much as 500 percent higher than nonprofits (Gail large electronic library of journals at a substan- Yokote, 2003). The price difference between for- tial discount, in exchange for a promise not to profit and nonprofit academic journals is particu- cut print subscriptions (whose prices will in- larly striking, given that these journals are generally crease over time); in this sense print and elec- similar in format and editorial processes, and that tronic are bundled. Since the electronic library for-profit journals do not appear to be of higher becomes much less expensive when ordered in quality (Theodore C. Bergstrom, 2001 p. 183). quantity, there is likewise bundling across elec- Increased concentration provides one possible tronic journals. explanation for why prices of for-profit journals Bundling can be seen as a device that erects are so high. To take one example, measured by a strategic barrier to entry. At a simple level of revenue, in 2001 Elsevier Science had a 22.9- analysis, the Big Deal contracts leave libraries percent share of the Science, Technology, and few budgetary dollars with which to purchase Medicine (STM) industry, with Kluwer at 11.7 journals from new entrants. Looking one level percent, and Thomson at 10.7 percent.1 But con- deeper, we see that bundling entails average centration offers at best only a partial answer. prices that exceed marginal prices, and this cre- Bundling offers another, potentially more signifi- ates a barrier to entry if entrants compete with cant explanation, particularly for recent increases. the marginal journal. Other things equal, bun- The prices of for-profit journals could not be dling practices are likely to be anticompetitive high and increasing without significant struc- to the extent that they allow for the maintenance tural barriers to entry. Recently, however, a new of supracompetitive average prices that limit usage of academic journals by scholars and/or distort library choices between journals and monographs and books. There are, however, † Discussants: V. Kerry Smith, North Carolina State University; Robert Hall, Stanford University. pro-competitive benefits associated with bundling. Recent deals have provided scholars with * School of Law and Department of Economics, Uni- versity of California, Berkeley, CA 94720. This research extra access to journals; moreover, when elec- is supported by a grant from the Andrew Mellon Foun- tronic databases contain journals not included in dation. The authors thank Catherine Candee, Mary Case, the libraries’ print collections, the collections Beverly French, Michael Keller, Mark McCabe, Ann expand. Okerson, Asunta Pisani, and Milt Ternberg for their helpful advice. Finding an economic approach that analyzes 1 Figures are based on data from Outsell: 70 percent of a range of bundling practices and evaluates worldwide STM revenue accrues to commercial publishers. them by appropriately balancing benefits and costs 441 442 AEA PAPERS AND PROCEEDINGS MAY 2005 offers a challenge to industrial-organization econ- status quo and the value of retaining existing omists.2 Our goal is to provide some initial per- investments help to explain the relatively in- spectives and to provoke further commentary. elastic nature of libraries’ demands for journals. A final barrier to entry is the substantial share I. Structural Barriers to Entry of a journal’s costs that are first-copy costs. The cost of recruiting, judging, reviewing, editing, The largest structural barrier to competing copyediting, and typesetting articles is essen- with an established journal is that journal’s rep- tially independent of the number of copies that utation. An established journal is a “coordina- will be circulated. Because journals can take tion equilibrium.” What makes a journal years to mature, the problem of achieving min- valuable is the simultaneous consensus of au- imal viable scale, let alone minimum efficient thors, reviewers, editors, libraries, readers, ten- scale, is substantial. Electronic publishing can ure committees, and indexing services that the change this landscape significantly, but the journal is of high quality. If all the actors were structural costs of entry remain substantial. It is to simultaneously conclude that an established not surprising therefore that there are few en- journal was low in quality and that a new jour- trants, most of which fill narrow niches, and nal was of higher quality, the quality ordering of most of which are subsidized by one means or the journals would be reversed, and with it their another. values. Because of the variety of actors that are required to reach a new equilibrium, the likeli- II. Bundling as a Strategic Barrier to Entry hood that such a reversal will occur and that a new entrant will supplant an existing journal is very small. A. The Big Deal: Profiting from Price Another barrier to entry is the libraries’ rel- Discrimination while Creating atively high switching costs and associated in- an Entry Barrier ertia. A library that has 40 years of back issues of a journal and is considering whether to buy While specifics of the “Big Deal” vary, a another year or to subscribe to a new or differ- typical offer would involve a multiyear contract ent journal, faces a different decision than if it in which the library agrees to keep its existing were choosing between the two journals de base of print subscriptions (with the base price novo. Unless the library’s journal is clearly increasing annually).3 The library pays a sur- dominated by a replacement journal, the library charge for access to an electronic journal data- will not shift its holdings for fear of devaluing base. A major sticking point for libraries has the existing collection. been their inability to glean substantial savings The inability to effectively monitor faculty by canceling print titles once they sign up for use is another problem. If monitoring were ef- the Big Deal. Libraries may get no savings from fective and libraries could both respond more a cancellation, or they may get limited savings accurately to faculty demands and more expe- that fall far short of the individual list prices the ditiously inform faculty about pricing issues library originally paid for the print journals. and the university’s need for faculty not to give The bundling of print and digital representa- their copyrights to journal publishers, it is quite tions of journals involves price discrimination. possible that libraries’ demands for journals In other markets, bundling often takes the form would become more elastic and that barriers to of second-degree discrimination, in which case entry would be reduced. Unfortunately, librari- buyers are offered the same pricing schedule but ans currently have limited information about the choose (often based upon unobservable demand usage of print journals. The stickiness of the characteristics) what to buy. Because some bun- dles have higher implicit per-unit prices or higher profit margins, this is viewed as a form 2 In Edlin and Rubinfeld (2004), we sought to identify and then provide an evaluation of the antitrust issues that might be raised either in private antitrust actions or in U.S. government non-merger investigations. The paper also pro- 3 Elsevier offers several Big Deal options (see Edlin and vides detailed descriptions of Big Deal contracts. Rubinfeld, 2004). VOL. 95 NO. 2 COMPETITION POLICY FOR JOURNALS 443 of indirect discrimination. The Big Deal bun- share of library budgets is committed in long- dling options are somewhat different, however; term contracts with existing publishers. There the discrimination is direct and is most appro- are indications that the Big Deal is hindering priately viewed as an approximation of third- entry. Librarians who have signed up for the degree price discrimination. The price that a Big Deal say that they would spend more buyer is quoted depends upon that buyer’s ob- money for journals from smaller and alternative servable characteristics (i.e., on the subscription publishers if they could achieve proportionate base prior to the introduction of Big Deal bun- savings from reductions at large publishers. dling). Moreover, the price of the Big Deal is often individually negotiated with a given library or with groups of libraries, which offers B. Bundling and Price Discrimination: further opportunities for publishers to price Where Do We Go from Here? based on individual characteristics. The result is that the bundled price reveals both a lower Bundling practices raise important questions bound and a reasonable approximation to each for industrial-organization economists. How are library’s willingness to pay for the bundle of these practices likely to evolve over time as journals as a whole. publishers learn more about libraries’ demands? A publisher could not do nearly as well at What is the best way to characterize the partic- discerning a library’s willingness to pay for a ular form of price discrimination at issue here? given journal, as it does for the bundle of jour- How are publishers’ pricing strategies likely to nals. The value of an individual journal to a change as libraries change their acquisition given school is highly ideosynchratic and can strategies? depend upon the preferences of a single vocal The bundling of academic journals has contin- faculty member. Whenever an incumbent pub- ually evolved. As this paper is being written, it is lisher misprices and loses a sale by pricing too probably accurate to characterize most “Big high for a school to buy, an opportunity is Deals” as having flipped from being print bundled created for entrants. Bundling takes advantage with electronic to electronic bundled with print. of the law of large numbers to limit such pricing Under the “electronic-bundled-with-print” sce- “inaccuracies” and with them the opportunities nario, the bulk of payments are for electronic for entrants. access, and these payments are required to buy This pattern of third-degree price discrimina- “cheap” print journals. The decision of a number tion is widespread. A survey of fellow librarians of major university libraries to opt not to enter into several years ago by Kenneth Frazier (2001) a Big Deal arrangement has encouraged publish- found that 66 percent had licensed Big Deals ers to put forward an even wider variety of bun- from the Academic Press and 60 percent from dling arrangements to achieve their goals. Library Elsevier (this was prior to Elsevier’s acquisition consortia have also become more creative in their of Academic Press). However, some schools thinking about how to respond to publishers’ bun- such as Stanford, Harvard, and Cornell have dling arrangements.4 shunned the Big Deal and have chosen to buy As negotiations continue from year to year, it electronic subscriptions a` la carte at dramati- is quite possible that publishers will be con- cally higher prices than print. vinced to break the link between print and elec- The immediate effect of the introduction of tronic journals. There is no reason to expect, Big Deal arrangements has been to move com- however, that bundling itself will disappear. It is petition from individual journals to large bun- crucial therefore to thoroughly understand the dles of journals. This move has created a very nature of the price discrimination that underlies substantial strategic barrier to entry into markets various types of bundling. for journals. Apart from the necessity of achieving viable scale in an industry where founding even a single journal requires the daunting task 4 See Jeffrey N. Gatten (2004), who proposes that con- of creating a new coordination equilibrium, a sortia monitor joint usage and utilize the information to new entrant needs to compete for library sub- negotiate reductions in Big Deal fees as members discon- scriptions in an industry where a substantial tinue low-use titles. 444 AEA PAPERS AND PROCEEDINGS MAY 2005

An important aspect of bundling is the effort buys Elsevier’s whole bundle, it may not have by a firm with market power to charge higher sufficient resources left to subscribe to the new prices by bundling substitutes. Consider a world journal. where a representative library purchases one The situation is similar for libraries that come copy of journal(s) A from firm 1 and one copy up for renewal of the Big Deal. A library sub- of journal(s) B from firm 2 and the journals or scribing to the complete Elsevier package can sets of journals A and B are substitutes so that cancel all 1,800 Elsevier titles and buy a`la the value of adding B to the collection falls if A carte. If the alternative publisher is only offer- is already held and vice versa. Even after a ing a single new journal, it is unlikely the li- merger, if the merged firm is restricted to sell A brary would cancel the Elsevier titles to get the and B separately (as Time-Warner and Turner new journal. Entry is certainly more difficult, were required to do with cable packages), and if but should the publishers’ behavior be deemed it desires to sell both, then the merged firm will to be exclusionary, or should the publisher be not be able to raise price. Bundling, however, seen as competing on the merits? allows it to raise price. By bundling A and B, We believe that bundling is a good candidate the merged firm can effectively stop A and B to be judged exclusionary and anticompetitive. from competing with each other, which substi- Excluding entrants by charging low prices is tutes will otherwise do even when sold by the generally considered competition on the merits: same firm. it is favored from a public-policy vantage be- Our initial analysis suggests that, in this spe- cause customers gain from the low prices. On cial case where libraries are homogeneous, bun- the other hand, because the total bundle price is dling does not serve as a barrier to entry. so much higher than the sum of the marginal Although it allows firms to charge higher prices, prices, Big Deal bundling excludes entrants access is not restricted, and competition appears without providing this kind of benefit to buyers. to be on the merits. Consider a new journal A complete answer to the question of whether which is a perfect substitute for journal A, but Big Deal bundling is exclusionary rests in part can be produced at lower cost. In this special on whether entry of new journals would be world, this journal will in fact successfully com- substantially easier if publishers did not bundle pete with journal A, and the merged firm will and only sold journals on an individual basis. cease to produce journal A. Our hypothesis is The answer presumably depends in part on the that if one considers a world where schools vary decisions of librarians as to whether and to what in demand (recalling the price-discrimination extent they would allocate more funds toward discussion above), then both efficiency-oriented new and alternative publications if they could and anticompetitive motivations for bundling achieve proportionate savings from cancelled will be important. subscriptions. We suggest that the bundling of a journal III. Is Bundling Harmful to Competition? publisher with monopoly power is exclusionary and anticompetitive if there exists a set of hy- A. Exclusionary Behavior pothetical new journals such that: (a) if each journal competes alone in the current market- The Big Deal and other forms of bundling place, it will likely gain limited acceptance and appear to be exclusionary in that they unneces- subscriptions and not thrive; (b) each journal sarily impair the ability of entrants to compete. could gain substantial market penetration and Because new entrants must compete at the mar- thrive if the monopolist did not bundle; (c) gin in the beginning, and because the bundle together, all of the journals could compete ef- entails a low marginal price, an independent fectively as a bundle against the monopolist’s publisher might gain few subscriptions even if it Big Deal bundle; and (d) it is substantially less creates a new journal with better articles than an likely that these new journals will be founded Elsevier journal that has a higher list (and av- and thrive if the monopoly sells its Big Deal erage) price. With bundling, cutting the Elsevier bundle. journal garners little savings that could be used If these conditions are satisfied, then what to buy the entrant’s journal, and if the library prevents entry of superior journals is not the VOL. 95 NO. 2 COMPETITION POLICY FOR JOURNALS 445 monopolist’s high quality or low-cost service, tions that choose a` la carte pricing with those but instead the strategies it chooses to sell them. that have the comprehensive Big Deal, but in- Hence, the publisher is not competing on the terpreting this comparison is complicated by merits. Condition (d) prevents the test from selection effects. Even if one could accurately being over-inclusive: if the preexisting entry determine how much extra research is made barriers were sufficient even without the Big possible by the Big Deal, putting a dollar value Deal bundle to preclude entry, then condition on reading journals that are not worth sub- (d) is not satisfied. scribing to on an a` la carte basis is inherently difficult. Likewise, it is difficult to gauge the extra benefit that desktop access provides to B. Pro-Competitive Benefits journals that would be on library shelves, but would not have electronic desktop access absent Recent bundling practices have allowed bundling. Moreover, if one is to attribute pro- many universities desktop access to journals competitive benefits to the bundling associated that otherwise could only be accessed by some- with the Big Deal, it is important to evaluate the one going to the library stacks. Some universi- extent to which those benefits would be ties have also been able to expand their access achieved without the Big Deal. If the non-Big to journals that they could not previously afford. Deal world were one in which Elsevier and What is the value of this potential efficiency other publishers offered more competitively benefit? Does it outweigh the possible costs of priced access to electronic subscriptions, many exclusion? Is there a substantially less exclu- of the apparent pro-competitive benefits might sionary way to achieve it? These questions do disappear. not have easy answers, in part because the issues raise inherent conflicts between the static efficiency gains associated with bundling and IV. Concluding Remarks the dynamic efficiency losses associated with a lack of additional entry. Bundling might also To summarize briefly, journal prices have promote large publishers to introduce more increased enormously as publishers have recog- journals over time to the extent that the price nized the market power they derive from having discrimination in bundling allows these publish- succeeded in a grand reputational game—a ers to capture more of the incremental surplus monster coordination equilibrium of authors, from extra journals (Doh-Shin Jeon and Djo- editors, reviewers, librarians, and indexers. menico Menicucci, 2003). Electronic publishing brought with it some- For-profit journal publishers can point to the what lower incremental costs (a small reduction likelihood of expanded use of existing journals in one barrier to entry) but also a bundled pric- under the Big Deal, both because electronic ing strategy that creates a substantial barrier to articles are easier to access than their print entry. New journals must to some extent com- counterparts and because the library subscribes pete with the bundle rather than with individual to more journals than it otherwise would. With journals within the bundle. At the same time, greater accessibility and lower search costs, this the bundling has benefits as it can increase usage could be substantial. The value of this access. usage is very hard to know, however. These electronic views and downloads of journals that would not be subscribed to with a` la carte pric- REFERENCES ing may substitute for time spent reading articles that would be subscribed to with a` la carte Bergstrom, Theodore C. “Free Labor for Costly pricing. Additionally, the value of this extra use Journals?” Journal of Economic Perspec- may accrue to the publishers in the form of tives, 2001, 15(4), pp. 183–98. higher prices, so this “efficiency” may not ben- Edlin, Aaron S. and Rubinfeld, Daniel L. “Exclu- efit users. sion or Efficient Pricing? The ‘Big-Deal’ One might attempt to gauge this usage sub- Bundling of Academic Journals.” Antitrust stitution by comparing usage patterns at institu- Law Journal, 2004, 72(1), pp. 119–57. 446 AEA PAPERS AND PROCEEDINGS MAY 2005

Frazier, Kenneth. “The Librarians’ Dilemma: Jeon, Doh-Shin and Menicucci, Djomenico. “Bun- Contemplating the Costs of the ‘Big Deal’ ”, dling Electronic Journals and Competition D-Lib Magazine, 2001, 7(3) [online: ͗http:// Among Publishers.” Unpublished manu- dx.doi.org/10.1045/march2001-frazier͘]. script, Universitat Pompeu Fabra, 21 April Gatten, Jeffrey N. “An Orderly Retreat from the 2003 [Available online: ͗http://www.ios.neu. Big Deal: Is it Possible for Consortia?” D-Lib edu/papers/s3i3.pdf͘]. Magazine, 2004, 10(10) [Available online: Yokote, Gail. “Faculty and the Scholarly Journal Pub- ͗http://www.dlib.org/dlib/october04/gatten/ lication Crisis.” Unpublished manuscript, Uni- 10gatten.html͘]. versity of California–Davis, 4 December 2003.