Untitled-6 March 1998114 College & Research Libraries Restructuring Serials Management to Generate New Resources and Services—With Commentaries on Restructurings at Three Institutions Charles A. Schwartz Restructuring serials management along access-versus-ownership lines does not solve the serials crisis, but it does alleviate inflationary pres­ sures and has other significant outcomes. The main economic outcome is the cancellation of low-use, high-cost-per-use titles to create a large pool of savings for reinvestment in new resources and services. The main service outcome is a fully subsidized, unmediated document de­ livery system—for a fraction of the cost savings. The main political out­ come is a resolution of the “tragedy of the commons” phenomenon, which tends to dominate library–faculty relations based on conventional seri­ als management. eople can define problems in ways that overwhelm their ability to do anything about them. Indeed, the serials crisis generally has been thought to be a sys­ tems-level economic problem having no workable solution. However, it is prima­ rily—on the local (real-world) level—a po­ litical matter akin to the “tragedy of the commons” phenomenon which can be al­ leviated with significant other outcomes. The “tragedy” is that the faculty have no reason to cooperate with conventional se­ rial cancellation projects, which amount to a perpetual budget-slashing act that leaves all stakehol der groups progressively worse off. Figuring out a more rational model in which all stakeholder groups come out ahead, economically and politi­ cally, is the essential problem. Variations of a new service model based on the restructuring of serials man­ agement along access-versus-ownership lines have been developed by a small number of academic libraries. The key factor is a fairly comprehensive serials cancellation project to generate a large pool of savings for reinvestment in new resources and services. The principal added service—the linchpin of the model—is a fully subsidized, unmediated document delivery system. Interest in this model has grown on the Internet and at conferences, but accounts of collections actually being restructured are rare.1 This article seeks to help fill that gap in the literature by providing a case study of the model at the University of Massachusetts- Boston (UM-B) and by presenting com­ mentaries from library managers at three Charles A. Schwartz is Assistant Director for Collection Management and Technical Services in Healey Library at the University of Massachusetts-Boston; e-mail: tony@delphinus.lib.umb.edu. Commentaries: 114 mailto:tony@delphinus.lib.umb.edu 1986 5 1987 7 1988 43 1989 35 1990 35 1991 28 1992 22 1993 15 1994 13 1995 8 1996 10 Restructuring Serials Management 115 FIGURE 1 The Rise and Decline of Publications on the Serials Crisis 50 45 40 35 30 25 20 15 10 5 0 1984 1986 1988 1990 1992 1994 1996 1998 other institutions on their respective ex­ periences. Restructuring serials manage­ ment at UM-B in 1996 was a three-part plan: • relying on faculty decisions, cancel­ ing hundreds of low-use, high-cost-per­ use (or other peripheral) serials for cost savings in the range of a quarter of a mil­ lion dollars; • reinvesting about 20 percent of the savings in an UnCover system that en­ ables authorized users to order, from their home or office, any article from a serial not in the library’s collection and to re­ ceive it, free of charge, usually within twenty-four hours; • reinvesting another 10 percent (or so) of savings in new serials and net- worked resources. This study is developed in six parts. The first part is an overview of the litera­ ture. The second contrasts serial cancel­ lation decision models. The third de­ scribes UM-B’s economic methods and outcomes. The fourth focuses on the reso­ lution of the “tragedy of the commons” phenomenon. The fifth provides a brief summary of the model’s strengths and limitations. And the last part presents the commentaries. Literature Overview In the main, the litera­ ture on the serials crisis consists of two types of studies. One type fo­ cuses on the oligopolistic structure of this publica­ tion market (in which a few producers dominate the scale of prices). None of those systems-level studies provides any real solution for making serials management more effective.2 Instead, they have called over the years for a variety of grand reforms, includ­ ing a general freeze on serial budgets, a fed­ eral subsidy for the scholarly commu­ nication system, a letter campaign against “greedy” publishers, a steep hike in photocopy charges to force the demand for private journal subscrip­ tions, and the abolition of the academic rewards system underlying publication proliferation.3–8 The bulk of the literature—more than 200 publications—is narrow and descrip­ tive, detailing inflation patterns in sub­ ject fields or announcing results of serial cancellation projects. Such reports tend to reiterate, in a line or two, the same sys­ tems-level proposals for some restruc­ turing of the serials market. The sharp decline of publications on the serials crisis this decade suggests that our pro­ f e s s i o n h a s t i r e d o f t h e u s u a l a c ­ counts—and perhaps that it has acqui­ esced to spiraling inflation (a specter once termed the “doomsday library ma­ chine”). Figure 1 shows the rise and de­ cline of publications on the serials crisis between 1986 and 1996. Decision Models No serial cancellation decision model fits all parts of the collection or applies to both restructuring projects and normal times. 116 College & Research Libraries March 1998 Cost-per-use analysis is superior to the conventional prices & quotas method for a comprehensive cancellation project in­ volving the sciences. As discussed below, a variant of cost-per-use analysis may be more expedient to cope with inflation in the sciences when restructuring is not in process, whereas the prices & quotas method seems inevitable for the social sci­ ences and humanities. Prices & quotas simply aims to expe­ dite a cancellation project by targeting the most expensive titles: Dollar quotas force bibliographers and faculty to factor the cost of a journal into the overall benefit that will be derived from it, to realize that the wisest choice may some­ times be to cut the ’best’ title on the list if it is by far the most expensive. It leads them, in short, to find a few titles that will go a long ways to­ ward getting it over with.9 An opposing model—the focus of this study—is cost-per-use analysis. Although the figures are easily computed with spreadsheets, the usage (reshelving) study is laborious and takes two semes­ ters to accumulate reliable data. Nonethe­ less, such analysis produces both inter­ esting micro-level findings (on a title-by-title basis) and persuasive macro- level indicators of the benefits to accrue from restructuring the collection. Cost-per-use analysis involves dual criteria: low use coupled with high cost per use (cost per use being the ratio of subscription price to usage).10 “High” cost per use is an estimated threshold, or tip­ ping point, at which ownership becomes much more expensive than access, whereas “low” use is a somewhat arbi­ trary range (usually zero to five). In ac­ tual practice, this kind of analysis is more varied and looser than the dual frame­ work. A survey (by this author) of fifty- three academic libraries that engage in access-versus-ownership decision-mak­ ing found a widespread tendency to un­ couple the criteria, putting little empha­ sis on low use while focusing on high cost per use.11 Thus, very expensive serials are tracked and, when cancellations become necessary to offset inflation, usage figures somewhat above the low-use line are dis­ regarded. Of course, to treat the low-use criterion as inviolable would impede bud­ get balancing. Yet, focusing on the most expensive titles has the ironic tendency of making this approach quite similar to the prices & quotas method it was in­ tended to replace. Cost-per-use analysis is superior to the conventional prices & quotas method for a comprehensive cancellation project involving the sciences. When the purpose of a cancellation project shifts from incremental budget cuts to a comprehensive restructuring of the collection, the criteria of low use and high cost per use must be accorded fairly equal status for the analysis to have a co­ herent structure. Roughly 80 percent of science serials have a cost per use in the hundreds of dollars, so it would be eco­ nomically rational to cancel the greater part of the science collection.12 For a balanced outcome, the low-use criterion comes into play—and that turns the analysis around, with about 20 percent of the science col­ lection then being targeted. Although the dual criteria make the analysis coherent, a dilemma arises for the library. On the one hand, coherence would last only as long as the low-use criterion was treated as a metadecision that no serials over that boundary line would be targeted. Coherence would thus rob the library of its title-by-title discre­ tion and flexibility. On the other hand, if the low-use criterion was relaxed by the library’s targeting some titles that were over the boundary line, there would be no logical stopping point because the great http:collection.12 http:usage).10 Restructuring Serials Management 117 TABLE 1 Access-versus-Ownership for UM-B, at a Glance Average Average Average Number Uses Cost per Use UnCover All science serials 19 $65 Range of use 296 low zero to 5 $537 $13 Subset of 255 marginal zero to 3 $605 If all 296 low-use, high-cost-per-use serials had been shifted from ownershipto access the previous year, total document delivery charges would have been about $3,000, less than two percent of total costs of $215,000. average for low- use, high-cost-per­ use titles ($537) or the subset of ex­ tremely low-use titles ($605).13 With nearly 300 titles in this pool of poten­ tial cost savings, the library could well afford to let the academic de­ partments pick and majority of science serials conceivably could be cut on strictly economic grounds. Thus, some mechanism is necessary to keep cost-per-use analysis out of the kinds of library–faculty frays that arise with the prices & quotas method. For a large cancellation project involving scores of potentially distrustful stakeholder groups on campus, coherence is no mere nicety. The way to resolve this dilemma of decision-making flexibility versus decision- model coherence is for the library to delegate micromanagement of the title-by-title de­ cisions to the academic departments. They can put varying emphases on the different criteria and consider the whole process to be quite reasonable. The library’s economic interests are secure because the overall cost savings will be substantial regardless of particular title decisions. UM-B’s Economic Methods and Outcomes UM-B’s resolve to restructure serials man­ agement began with a single striking sta­ tistic. As shown in table 1, the provision of document delivery for every low-use, high-cost-per-use serial in the sciences would cost about two percent of their to­ tal subscription costs—a savings of well over $200,000 a year. Another telling sta­ tistic in table 1 shows that a large cancella­ tion project was bound to generate substan­ tial savings, given the ninefold difference between the average cost-per-use of the sci­ ence collection as a whole ($65) and the choose within the pool. "Good Enough" Aggregate Projections14 Projected cancellations at UM-B ranged from $200,000 to $225,000 for the sciences alone. Any such outcome was “good enough” because the share of savings to be reinvested in document delivery (esti­ mated at $40,000 a year, including the $10,000 UnCover fee for a customized system) varied only five points (from 17 to 20%). The actual cancellation total (in­ cluding nonscience fields) was $227,000. With the additional reinvestments (115 new serials and some databases), the sav­ ings amounted to three-quarters of former ownership costs and offset future inflationary pressures by a third. On that scale, the somewhat arbitrary values that were selected to define the low-use criterion and the access-versus­ ownership threshold hardly mattered. UM-B had set the low-use criterion at five but discovered that 86 percent of low-use serials were in the zero-to-three range. As for the access-versus-ownership thresh­ old value ($18), it could have been doubled without lowering the share of cost savings by more than two percent­ age points. The Relative Insignificance of the Price Factor in the Sciences Subscription price was a relatively insig­ nificant decision factor compared to cost per use in the sciences. Table 2, based on five departments that accounted for 70 118 College & Research Libraries March 1998 TABLE 2 Serial Price as an Insignificant Factor Relative to Cost-per-Use (1) (2) (3) (4) (5) (6) (7) (8) Average Price Average CPU Low Use, Per Standard Extreme Double Whammies Per Low-Use High C.P.U. Field Deviation Price Whammies* Cut Field Titles Biology $531 $669 $868 $1,537 8 6 $35 $261 Chemistry 871 1199 1402 2601 2 1 106 686 Environmental 593 672 748 1420 3 1 52 465 sciences Mathematics 548 514 536 1050 16 6 143 548 & computer science Physics 1376 1489 2126 3615 2 1 180 876 31 15 Weighted avgs $710 $827 $1026 $1853 $90 $504 * Double Whammy = A title one standard deviation in price for that field (usually triple the average price) and low in use. percent of all cost savings, outlines the main reasons. As shown in columns 1 and 2, the average price of low-use, high-cost­ per-use titles was actually lower than the average price in four of the five fields. Next, consider the extreme case of double whammy serials, those of very high cost and low use. The “extreme price” (column 4) threshold begins at one standard de­ viation above the average price, with most double whammies tripling the average price (due to big standard deviations). Of those thirty-one titles, the science depart­ ments decided to keep half (columns 5 and 6). Because any librarian with a price & quotas perspective would have targeted all thirty-one, the case of the double whammies might seem a worst-case sce­ nario of academic departments’ broad discretion. Yet, the sixteen retained double whammies amounted to only $24,000—an amount that would have in­ creased total project savings just 10 per­ cent. With those five science departments’ cutting $170,000 overall, making a fuss over some titles they considered indis­ pensable would have been impolitic. Old-Style Review in the Social Sciences and Humanities Without usage data for the social sciences and humanities, UM-B’s serials review was more conventional. In proportional terms, the results were “good enough” for the universitywide process to be judged eq­ uitable: Cuts in the social sciences and hu­ manities amounted to 25 percent of those budgets, as opposed to an overall rate of 34 percent in the sciences. In absolute terms, of course, the results were disparate. The social science and humanities departments cut almost twice as many titles (313, as opposed to 188 in the sciences) but repre­ sented only a fifth ($52,000) of aggregate savings. The different price structures were most reflected in extreme price serials (those at least one standard deviation above the average price in a field). Social science and humanities departments cut five times as many such titles (73 of 200) as did science departments (15 of 31) but gen­ erated barely the same savings ($32,500, as opposed $30,000 in the sciences). However, extreme price title cuts accounted for two- thirds of the overall contribution of the social sciences and humanities to the project, mak­ ing the process politically equitable. Restructuring Serials Management 119 Resolving the “Tragedy of the Commons” The “tragedy” pertains to situations in which a limited resource (or public good) is shared by a number of users who, al­ though they must act on a basis of fair­ ness or equity to conserve that resource, have no immediate incentive to do so. Self-interested behavior eventually de­ pletes the resource, leaving everybody worse off. The textbook example of a meadow shared by too many herdsmen may be intuitively familiar: Each economically rational herds­ man [academic department] seeks to get the most benefit possible of the common pasture [serials bud­ get] by maintaining a sizeable herd [serials list]. A day of reckoning comes when the commons starts to be depleted. In agrarian society, cows die off [on campus, serials are cut].15 Conventional serials management, eco­ nomically and politically, is quite conser­ vative. A continuous series of incremen­ tal cancellation projects every few years amounts to a model of “muddling through.” With the new service model, however, the library’s role changes from micromanaging serial cuts to orchestrat­ ing what amounts to a paradigm shift for the faculty. This new role depends on all academic departments being assured that the library will exercise its responsibility of enforcing the equity of the cancellation project in order to avoid the “tragedy of the commons” phenomenon. Restructuring Seen as an Intelligent Move An intelligent move is often thought to require clear-cut outcomes. In a compre­ hensive serials review, however, particu­ lar outcomes (titles cut, especially double whammies) are far less important than the character of the process itself. As we have seen, access-versus-ownership decision factors are neither precise nor prescrip­ tive. Although the intelligence of restruc­ turing is as broad and concrete as the idea of doing things in cost-effective ways, it is primarily local and symbolic, based on the library’s decision to rely on the discretion of the faculty. That initial decision goes a long way toward resolving both the “tragedy of the commons” phenomenon and the dilemma of a flexible-versus-coherent decision process. So much depends on who is taking the initiative! If the faculty is put in the usual position of defending their fields from the library’s targets, the restructuring will not appear to be an in­ herently intelligent move. As a case in point, consider the argument for the li­ brary to target all cuts: “The fear some faculty have of what those idiots in the library might do if left to their own de­ vices can be our most powerful source of leverage in motivating faculty participa­ tion.”16 With regard to numbers and statistics, the intelligence of restructuring is best conveyed with vivid information that is local, concrete, and easy to remember (see figure 2). On a general level, the kind of information shown in tables 1 and 2—the The social science and humanities departments cut almost twice as many titles (313, as opposed to 188 in the sciences) but represented only a fifth ($52,000) of aggregate savings. extreme differential between high and av­ erage costs per use, together with the com­ parison of overall access and ownership budgets—should be prominent in cam­ pus meetings on the new service model. Statistics on inflation, by contrast, are rela­ tively pallid and point in the wrong di­ rection—at the problem rather than the solution. Top-down Deliberation and Bottom-up Review Although it is crucial for the library to en­ gage the powerful science departments from the start, campus politics must ad­ N o te : T h e fi rs t se ri al ti tl e is i n b o ld f o n t to e m p h as iz e it s lo w u se . F o r th e o th er t it le s sh o w in g m o d er at e to h ig h u se , t h e C .P .U . f ac to r is n o t ta rg et ed . F ig ur e 2 P ar ti al S am pl e S pr ea ds he et 120 College & Research Libraries March 1998 F IG U R E 2 P ar ti al S am pl e Sp re ad sh ee t B IO L O G Y (C .P .U . = c os t p er u se ) M ea n $6 69 A ve ra ge C .P .U . i n fi el d $3 4 S ta nd ar d de vi at io n $8 68 N um be r of lo w -u se ti tl es 56 E xt re m e pr ic e $1 ,5 37 A ve ra ge C .P .U . o f hi gh C .P .U . t it le s $2 61 To ta l c os t o f lo w -u se jo ur na ls $2 9, 76 6 [S am pl e E nt ri es ] C o u n tr y C o st U se C P U M O L EC U L A R A N D C EL L U L A R B IO C H EM IS T R Y U S $2 ,7 28 .0 0 3 $9 09 .3 3 U nc ov er JO U R N A L O F C O M P A R A T IV E P H Y S IO L O G Y ( A & B ) G E $2 ,7 63 .0 0 19 $1 45 .4 2 C E L L A N D T IS S U E R E S E A R C H G E $2 ,8 51 .0 0 n ew U n co v er F E M S M IC R O B IO L O G Y N E $3 ,9 30 .0 0 14 $2 80 .7 1 U n co v er B IO C H IM IC A E T B IO P H Y S IC A A C T A N E $7 ,5 55 .0 0 81 $9 3. 27 N ot e: T he f ir st s er ia l ti tl e is i n bo ld f on t to e m ph as iz e it s lo w u se . F or t he o th er t it le s sh ow in g m od er at e to h ig h us e, t he C .P .U . f ac to r is n ot ta rg et ed . dress the concerns of all stakeholder groups. Top-down deliberation begins by enlisting the support of the provost, pro­ ceeds to lunches with the academic deans individually, and then shifts to group meetings with the aca­ demic department chairs of a given college or di­ vision. A group setting reinforces the chairs’ in­ terdependence for the achievement of equitable outcomes in the serials review (i.e., it brings to the fore the potential problem of “free riders,” who would not make their fair share of cuts). Bottom-up review, coming after the depart­ mental access-versus­ ownership decisions, is a university-wide “safety check” with the distribu­ tion of two lists of all cuts, one for the sciences and the other for the social sci­ ences and humanities. In this review, the library adjusts cross-departmen­ tal or interdisciplinary needs. By releasing chairs of the concern that they might cancel a title of im­ portance to another de­ partment, it both expe­ dites the decision pro­ cess and enlarges the number of cuts. It also frees the chairs from be­ ing personally involved in a dispute with a col­ league over some title; otherwise, chairs might reverse decisions, caus­ ing savings—and process equities—to unravel. Another kind of “safety check” is UnCover ’s monthly reports on all se­ rial requests and individual requestors. The data would enable the library to make reverse decisions, from access to ownership, for seri­ als in high demand. However, in UM-B’s experience, 70 percent of document deliv­ Restructuring Serials Management 121 eries have involved only one article from a particular serial, and less than 5 percent of serials involve more than four docu­ ment deliveries. As for overuse of the ser­ vice, we have asked only two people out of hundreds to “flag” more citations (for interlibrary loan) and not to “mark” so many (for document delivery).17 In UM- B’s safety check review as a whole, 10 per­ cent of title cuts (50) were appealed and about one in seven reinstated by the project head, affecting less than 3 percent of total savings. This review had a largely symbolic role: The faculty seemed less interested in getting their way than in a formal mechanism for having a say. Concluding Remarks The new service model for serials access has three main strengths: It provides for informed planning, improved relations with the faculty, and new resources and services at a fraction of cost savings. The model’s limitations also are summarized here. A final point of discussion is the speculative issue of why more libraries have not adopted this model. Model Strengths With respect to informed planning, the library can determine quite early the eco­ nomic prospects of a fairly comprehen­ sive serials cancellation project, the key factor being the large differential between average cost per use and high cost per use for hundreds of low-use titles.18 Improved relations with the faculty will accrue on several fronts. The library’s delegation of access-versus-ownership decisions to the faculty will make the cancellation project appear to be an inherently intelligent move. In addition, the faculty no longer will be in a defensive situation, which is the crux of disputes; and they will take the serials review quite seriously, having been given both interesting data on the use of the collection and the quid quo pro of new subscriptions and even databases along with the document delivery system. At the same time, the cancellation project will be an instrument of confidence build­ ing in the ability and resolve of the library to come to grips with a perennial finan­ cial and service problem. Model Limitations The new service model is not a resolu­ tion of the serials crisis; prolongation of serials inflation will require recurring cuts over the years. Also, because cost-per-use analysis is not suitable for the social sci­ ences and humanities, the approach taken by many of the fifty-three surveyed librar­ ies of doing usage counts for expensive titles in those fields looks to be the only feasible course. A third limitation of the service model is that fax delivery does not work for articles with images. Especially in the biological sciences, the library should have an additional, surface-mail document delivery service (e.g., the Brit­ ish Library Document Supply Centre). A few aspects of UM-B’s experience with the document delivery service war­ rant attention. The serials it owns, and articles costing more than $30, are blocked from the ordering process. All of the aca­ demic departments that rely on the ser­ vice were “early adopters”—that is, they began to use it within a few weeks of its availability—whereas this technology has not spread to some other departments even after a year of publicity. An unex­ pected finding was that over 90 percent of deliveries go to home or office fax ma­ chines (i.e., this system made for a “li­ brary without walls” rather than the li­ brary becoming a “print house”). Why Have Not More Libraries Restruc­ tured Serials Management? Does UM-B recommend the new service model? Not exactly; “recommend” is too strong a term. We suggest first a full un­ derstanding of all the work involved. The initial usage study is year long. Then, preparation of subject spreadsheets, even with vendor-supplied electronic data, may take someone working full-time a month to do. An updated holdings tape http:titles.18 http:delivery).17 122 College & Research Libraries March 1998 (for the document delivery system to block requests involving serials owned by the library) could be a stumbling block for some libraries. The essential point, however, is that serials management is not necessarily locked into “captive mar­ ket” economics and “tragedy of the com­ mons” politics. The main reason more libraries have not restructured may be that they have not had the striking experience of seeing for their own budget and collection the kind of economic and service projections in tables 1 and 2. The untenable prospects for “serials management as usual” are well known, yet, on a workaday level, still distant and abstract. Unless the extreme difference between the access and the ownership models with regard to low- use, high cost-per-use serials is based on local figures, it simply does not seem real enough to cause what amounts to a para­ digm shift for the library and its stake­ holder groups. There is considerable evidence from the behavioral sciences that decision mak­ ers who are committed to particular pro­ grams tend to discount, ignore, or attack information that is at odds with such pro­ grams.19 Thus, faculty may disparage se- rial-usage data until that issue is defused by the library’s delegation of title deci­ sions to the academic departments. Col­ lection managers may remain committed to the ownership model, even after their university administrations have man­ dated significant budget cuts. For that reason, a case study on restructuring probably will not change attitudes unless there is a predisposition to consider fun- The untenable prospects for “serials management as usual” are well known, yet, on a workaday level, still distant and abstract. damental change. A case study can be a cognitive map of the concepts, relations, and strategies of an organizational situa­ tion. However, a decision to restructure will likely require a preliminary analysis to see firsthand how prospective cost sav­ ings would generate new resources and services and alleviate inflationary pres­ sures. Notes 1. On restructuring a serials collection, see Charles A. Hamaker, “Re-Designing Serials Col­ lections,” Journal of Library Administration 20, no. 1 (1994): 37–47; David F. Kohl, “Revealing UnCover: Simple, Easy Article Delivery,” Online (May/June 1995): 53–60; Jane Kleiner and Charles A. Hamaker, “Libraries 2000: Transforming Libraries Using Document Delivery, Needs Assess­ ment, and Networked Resources,” College and Research Libraries 58 (July 1997): 355–74. 2. On the nil prospects of a restructuring of the serials market, see Charles A. Schwartz, “Scholarly Communication as a Loosely Coupled System: Reassessing Prospects for Structural Reform,” College & Research Libraries 55 (Mar. 1994): 101–17. 3. Association of Research Libraries, Report of the ARL Serials Prices Project (Washington, D.C.: ARL, 1989). 4. David W. Lewis, “Economics of the Scholarly Journal,” College & Research Libraries 50 (Nov. 1989): 674–88. 5. H. Craig Petersen, “The Economics of Economics Journals: A Statistical Analysis of Pric­ ing Practices by Publishers,” College & Research Libraries 53 (Mar. 1992): 176–81. 6. Bruce Kingma and Philip Eppard, “Journal Price Escalation and the Market for Informa­ tion: The Librarians’ Solution,” College & Research Libraries 53 (Nov. 1992): 523–35. 7. Anthony M. Cummings et al, University Libraries and Scholarly Communication (Washing­ ton, D.C.: ARL, 1992). 8. Michael Stoller, Robert Christopherson, and Michael Miranda, “The Economics of Profes­ sional Journal Pricing,” College & Research Libraries 57 (Jan. 1996): 9–21. 9. Paul Metz, “Thirteen Steps to Avoiding Bad Luck in a Serials Cancellation Project,” Jour­ nal of Academic Librarianship 18 (May 1992): 79. http:grams.19 Restructuring Serials Management 123 10. Dorothy Milne and Bill Tiffany, “A Cost-per-Use Method for Evaluating the Cost-Effec­ tiveness of Serials: A Detailed Discussion of Methodology,” Serials Review 17 (summer 1991): 7– 19; Eleanor A. Gossen and Suzanne Irving, “Ownership versus Access and Low-Use Periodical Titles,” Library Resources & Technical Services 39 (Jan. 1995): 43–52. 11. On December 12, 1995, I posted a request on two listserves (COLLDEV-L and PACS-L) for libraries that do cost-per-use analysis for access-versus-ownership decisions to identify them­ selves for a survey. From more than 60 responses, 53 completed the survey in spring 1996. 12. The economic rationality of canceling an entire science collection, making a complete shift to an access model, is not complete hyperbole; see Richard P. Widdicombe, “Eliminating All Journal Subscriptions Has Freed Our Customers to Seek the Information They Really Want and Need: The Result—More Access, Not Less,” Science & Technology Libraries 14, no. 1 (1993): 3–13. 13. For case studies of cost-per-use analysis, see: Dorothy Milne and Bill Tiffany, “A Survey of the Cost-Effectiveness of Serials: A Cost-per-Use Method and Its Results,” Serials Librarian 19, no. 3/4 (1991): 137–49; Sue Beardman, “The Cost-Effectiveness of Access versus Ownership: A Re­ port on the Virtual Library Project at the University of Western Australia Library,” Australian Library Review 13 (May 1996): 173–81; Anthony W. Ferguson and Kathleen Kehoe, “Access vs. Ownership: What Is Most Cost Effective in the Sciences?” Journal of Library Administration 19, no. 2 (1993): 89–99; Diane E. Reuss, “Changing Strategies for Information Delivery: Theory into Prac­ tice,” Journal of Interlibrary Loan, Document Delivery & Information Supply 5, no. 4 (1995): 43–51; Tina E. Chrzastowski and Mary A. Anthes, “Seeking the 99% Chemistry Library: Extending the Serial Collection through the Use of Decentralized Document Delivery,” Library Acquisitions: Practice & Theory 19 (summer 1995): 141–52; Gossen and Irving, “Ownership versus Access and Low-Use Periodical Titles.” 14. The concept of “good enough” estimates or projections has a distinguished pedigree as the basis of Nobel prizewinner Herbert A. Simon’s theory of the “satisficing” decision model. Herbert A. Simon, Reason in Human Affairs (Stanford, Calif.: Stanford Univ. Pr., 1983), 85. 15. Adapted from Garrett Hardin, “The Tragedy of the Commons,” Science 162 (Dec. 13, 1968): 1,243–48. 16. Metz, “Thirteen Steps to Avoiding Bad Luck in a Serials Collection Project,” 77. 17. Conventional wisdom holds that free services to the typical user are devalued to the point of being abused—the “free rider” problem in the “tragedy of the commons.” However, some recent research suggests that this problem is more theoretical than real, that users of free services do conserve them when they discuss them. For a brief overview, see Jeffrey Pfeffer, New Direc­ tions for Organization Theory: Problems and Prospects (New York: Oxford Univ. Pr., 1997), 75–77. 18. Only a few cost-per-use studies have identified this differential; at Albany-SUNY, it was tenfold for the sciences (Gossen and Irving, “Ownership versus Access,” 46); at Columbia Uni­ versity, it was eighteenfold, a figure actually deflated by a very high document delivery estimate in the early 1990s of $39 (Ferguson and Kehoe, “Access vs. Ownership,” 94). 19. Sara Kiesler and Lee Sproull, “Managerial Response to Changing Environments: Perspec­ tives on Problem Sensing from Social Cognition,” Administrative Science Quarterly 27 Dec. 1982): 548–70. March 1998 124 College & Research Libraries Commentaries on the New Model at Other Libraries No single study can be simultaneously accurate (scientific), general (theo­ retical), and simple (practical). To secure any two of these virtues auto­ matically sacrifices the third one. The case study by Charles A. Schwartz on pages 115–124 of a new serials model is simple and accurate, thus its generality is inevitably doubtful. The following commentaries make the case study more general and accurate but less simple. Those are inevitable trade-offs. Colorado State University Joel Rutstein Colorado State University (CSO) had the original inspiration for what would be­ come the UnCover SUMO (subsidized unmediated ordering) document delivery service. That formative idea was not the product of a declining economic situation but, rather, an improving one. After a $100,000 serials cancellation project in 1992, we had an abundant materials bud­ get that was ranked as the number one priority in the university’s strategic plan. Why, then, did CSO propose in 1993 that the Colorado Association of Research Li­ braries invent SUMO? Simply, if the cur­ rent situation is healthy, what better time to experiment? Thus, SUMO developed out of a CSO pilot project to analyze behavioral patterns of library users in what would be a sub­ stantially new information environment. As pioneers in this area, we needed to dis­ cover how a cost-free document delivery service actually would be used, its finan­ cial ramifications for the library, its impact on interlibrary loan, and whether—despite being unmediated—it would require much staff intervention. In 1994, CSO became the first library to offer UnCover SUMO. Undergraduates and residents of Fort Collins were in­ cluded in the authorized-user pool. Two controls were placed on the service: All titles owned by the library, and articles costing more than $25, were blocked from the subsidy. After the service had been in operation for two years (through 1996), the following conclusions were drawn about the nature of the service in CSO’s environment: � Interlibrary loan: There was no ap­ preciable decline in demand, in that ar­ ticles available through UnCover con­ tinue to be requested through ILL. � Overall cost: From $28,700 in FY1995, the cost rose to $40,900 in FY1996 and might plateau at about $45,000 in FY1997. One source of rising cost is copy­ right charges: They averaged $4.70 in FY1995, $5.74 in FY1996, and $6.50 in FY1997. The average document delivery charge was about $13 for FY1997. The other source of rising cost is spreading use of the service at CSO. � Serial access frequency: About half (48%) of all serials in CSO’s document delivery ex­ perience involve one article only, whereas 15 percent involve five or more requests. Joel S. Rutstein, formerly Collection Development Officer, is Coordinator of the Gifts Program in the William E. Morgan Library at Colorado State University; e-mail: jrutstein@manta.library.colostate.edu. 124 mailto:jrutstein@manta.library.colostate.edu Restructuring Serials Management 125 TABLE 1 User-Group Behaviors Group Faculty Staff Grad students Undergraduates UnCover Browsing 10% 6 43 41 UnCover Orders 19% 7 52 22 ILL 30% 10 50 10 A key finding is the greater use of UnCover by students, for both browsing and ordering, than by the faculty, who generally are conserva­ tive about altering their research methods. �Fill rate and delivery time: The fill rate was 87 percent, with the follow­ ing delivery times: one hour, 17 percent; twenty-four hours, 44 per­ � Access-versus-ownership costs: Of the 15 percent of serials involving five or more requests, one in six had document delivery charges greater than their sub­ scription costs (overall $3,179 versus $1,896), the other five had access charges far lower than subscription costs (overall $14,062 versus $126,597); on balance, a savings of $109,000. University of Kansas Melvin G. DeSart and Rachel Miller The restructuring experience of the Uni­ versity of Kansas (KU) varies from that of UM-B in three respects: KU’s serials review process was gradual and multiyear, as opposed to UM-B’s all-at­ once approach; KU’s document delivery service was not implemented in concert with a particular serials review, whereas UM-B used the prospect of document de­ livery as a strategy with faculty; and the access-versus-ownership decision models differed. Over the past ten years at KU, many high-cost, low-use serials were cut. In fact, KU’s implementation of a customized UnCover SUMO system in 1995 was driven largely by the cumulative effect of such cancellation projects. Whereas cent; two days, 21 percent; longer than two days, 18 percent. Overall, usage of the service reflects the library’s collection development al­ location formulae, which give the great­ est weights to graduate student needs and next to faculty needs. Continuing analyses of UnCover will be integrated into the larger planning for electronic in­ formation access. UM-B’s UnCover system was part and parcel of a single, fairly comprehensive project, KU’s UnCover implementation and subsequent serials restructuring were the main ingredients of a remedy for a long-term trend of diminishing serials ownership. By 1995, when KU implemented its customized SUMO service, there simply were not enough high-cost, low-use titles remaining in the KU science and tech­ nology collections to merit conducting the type of cost-per-use study done at UM-B. Rather, in the next cancellation project (spring 1996), we chose to fol­ low the so-called “LSU model” by ask­ ing each faculty member to provide the serials he or she needed on-site. That entire process, including cancellation decisions, took less than four months. The resulting science and technology serial cuts ranged from a few titles that Melvin G. DeSart is Engineering Librarian and Coordinator of Science Collection Department in the Spahr Engineering Library at the University of Kansas; e-mail: desart@ukans.edu. Rachel Miller is Head of Retrieval Services in the Watson Library at the University of Kansas; e-mail: miller@ukans.edu mailto:miller@ukans.edu mailto:desart@ukans.edu 126 College & Research Libraries March 1998 had not been on any faculty member ’s list to those that appeared on several lists but were consistently low ranked. Esti­ mated cost per use was another decision factor, particularly when similar titles were being compared, but the serials can­ celed ran the price gamut, from $50 to more than $5,000. In the decision-making process, we chose to place primary em­ phasis on faculty needs and local usage, within the inevitable limits of the materi­ als budget. Past and continuing serial cancella­ tions (creating more access demand), combined with growing awareness of the document delivery service, have caused at least a doubling of its initial cost ($25,000) during the past two years. Yet, the projected $50, 000 to $60,000 cost for FY1998—being just five to six percent of the $953,922 in former ownership costs New Jersey Institute of Technology James Robertson In early 1995, the New Jersey Institute of Technology (NJIT) library canceled 150 serials, all available through UnCover, for a savings of $100,000. Most of the savings were put in a document delivery fund that would support not only users’ un­ mediated orders but also a good deal of interlibrary loan (ILL) requests. Unlike UM-B, NJIT did not have serial-usage data to identify low-use, high-cost-per­ use titles. Instead, the library relied on faculty members’ own estimates (and bi­ ases). Later that fall, NJIT launched its UnCover SUMO system. Unlike most other universities that offer this service, NJIT allows its undergraduates to use it. Moreover, NJIT may have been unique in of the 2,854 serials which KU has canceled since FY1991—is a bargain, especially in terms of the access and service provided. We would underscore a point Tony Schwartz mentions: that creating a hold­ ings tape (for the system to block orders involving serials held on-site) is an im­ portant reason to understand all the work involved in establishing this new service model. Because UnCover matches only on an OCLC or an International Standard Se­ rial Number, the serials department had to add ISSNs to many records and even, when it was necessary to force-match titles that did not have either a usable OCLC number or ISSN, to create a special field in the bibliographic record to contain UnCover’s identifier. Overall, the holdings tape was a laborious problem that had to be resolved if the new service was to be as cost-effective as possible. initially offering the service without any restrictions (i.e., there was no blocking of articles costing over a certain amount, nor were the titles held on-site blocked). That decision reflected a dual concern for col­ lecting realistic data on user needs while avoiding inappropriate restrictions based on assumptions. Economic restrictions eventually proved necessary. Whereas in FY1996 di­ rect orders by users were $21,777 and ILL orders $20,652, in FY1997 direct orders were about $42,000 and ILL orders about $14,500. When projections showed UnCover expenditures outrunning avail­ able funds, the library put a block on titles held on-site and on articles costing more than $30. The library currently is work­ ing with UnCover to limit the total dol­ lars each member of a user class may spend before one must apply to the li­ brary for a new account. (Possible limits James Robertson is Director of the Architecture Library of the Robert W. Van Houton Library at the New Jersey Institute of Technology; e-mail: robertson@tesla.njit.edu. mailto:robertson@tesla.njit.edu would be $300 per undergraduate stu­ dent, $500 per staff member, and $1,000 per graduate student or faculty member.) UnCover provides a SUMO library monthly computer disk report with de­ tails on every order (e.g., user name, date and time of order and delivery, cost and bibliographic information). Library staff integrate all reports into a cumulative Microsoft Access database. Ideally, the UnCover system would allow the library to fine-tune the level of restrictions on individual users. The long- term solution is to reinstate the library in the order transaction—but as an auto­ mated process, rather than a manual one. The NJIT library has proposed d e v e l o p i n g c o m p u t e r m i d d l e w a r e ( s o f t w a r e ) w h i c h w o u l d re s i d e o n campus. Users would go through the m i d d l e w a r e t o a c c e s s U n C o v e r. When a user submits an order, the middleware would first check its files for restrictions, such as the user ’s cu­ mulative cost over a given time period, the cost of each article, whether the seri­ als involved are held on-site, and whether there was a duplicate order for that user. Restructuring Serials Management 127 If the middleware authorizes the order, it would pass the order to UnCover and update the customer ’s local order history file. This interaction would be transpar­ ent to the user. The middleware would be an improve­ ment over the current system in two ways: It would allow sophisticated, mul­ tivariate automated mediation; and it would enable the library staff to quickly monitor economic transactions and user behavior. Further, the middleware con­ ceivably could be developed to work with other document delivery vendors. Users would neither know nor care whether the requested article came from UnCover or, for example, UMI, CISTI, or IEEE. The middleware would make an automated decision based on cost efficiency and user satisfaction criteria. The NJIT library be­ lieves this is an important future direc­ tion in library service and invites collabo­ ration with libraries and vendors interested in such middleware. Further information on the NJIT library’s UnCover project is updated periodically on the library’s Web site at: http://www.njit.edu/njIT/Li­ brary/uncover/report.html. http://www.njit.edu/njIT/Li