by Karen Coyle, Digital Library Consultant
Preprint. First published in the Journal of Academic Librarianship, August, 2006, v. 32 n. 5
Libraries are somewhere in between the investment-conscious business and the fashion-conscious individual. Libraries have to justify their technology budgets to their boards or their institutional oversight committee. However, some amount of the expenditure on technology is not a matter of a measurable return but is simply the cost of continuing to be a player in a high technology information society. Any library that fails to keep abreast of technological change will be quickly seen as irrelevant in the larger information context. This isn't just a question of fashion: we truly function in an interlocking web of information services, and continued participation has its requirements.
Libraries do not usually perform rigorous analyses to justify their purchases of equipment. There is an assumption in the profession that libraries cannot calculate a return on investment because our services are intangible and cannot be measured. That's a shame, actually, because we may be losing an opportunity to show that libraries have achieved some impressive efficiencies through their adoption of technology.
The business calculation of Return on Investment (ROI)1 is based on a view of investment that is deeply embedded in the concepts of capitalism: you invest capital to create profit. Some economists consider the ROI to be an overly vague measure, and one that can easily by manipulated to show whatever outcome one wishes to put forth. Still, it's one of the few ways that we have to measure costs against benefits. There are many different things that ROI can measure, mainly because you can define "investment" and "return" to suit your situation. An investment can be paying for advertising to an outside agency, or training for your staff. You can invest money in equipment or you can invest your employees' labor in a project. Similarly, return is defined in terms of your goals. In businesses, return can be greater profits, or it can be getting a particular product placement. Return can be also relative, such as increasing sales as compared to a previous time, or decreasing the number of manufactured items that are rejected for defects. Because we can define our own concepts of investment and return it is possible that we can use ROI for library activities.
The business use of the term ROI implies certain sophisticated calculations of the value of an investment over time. I'd like to make clear that I am speaking of return on investment as a general concept of costs and benefits. This sense is used in contexts that are relevant to libraries and their service functions, such as when Jakob Nielsen refers to the return on investment of usability in a web site2 or in the concept of the social return on investment (SROI) that has been derived from the business model.3 SROI would be interesting for the libraries because it attempts to measure the impact that an organization has on society rather than the impact that the organization's activities have on its own bottom line. The calculation of SROI for academic libraries, however, is inherently tied to the value that the academic institution itself provides, making it particularly complex.
We can instead look at the comparative calculation of return on investment, that is, doing something better than before. Of all modern library functions, the cost of producing the library catalog is often challenged as being overly expensive, and with the recent projects to digitize large numbers of books from library holdings, some see the need for a catalog waning. While the actual value of the catalog itself is hard to measure, it seems intuitive that the use of technology has brought those costs down dramatically, and at the same time has resulted in an improved quality of the data. The question is, can we prove this?
We know that library catalogs are expensive to create and maintain, although this being an essential service of the library, foregoing the creation of a catalog is probably not an option that many libraries will consider. At least one author, Joe Matthews, has attempted to calculate the value of the library catalog,4 including the value of the information contained within it. Finding an absolute value for the catalog requires making use of some of the factors of a social return on investment, including the value of the knowledge contained in the library. This is a tall order, and one that most libraries have no need to undertake. Yet we can, at least in theory, compare the relative cost of creating the catalog using different methods.
We have experience in this area. As a matter of fact, the library community underwent a massive technology upgrade relating to catalogs and cataloging when it moved from cards and card catalogs to bibliographic services and online catalogs. Libraries invested heavily in what was really a collective leap of faith. Fortunately, the results were generally positive. If we knew then what we know today, we could have shown that our decisions to "go electronic" were based on a wise investment strategy, even if we didn't know it at the time.
In 1976 or so, libraries began using the services of bibliographic utilities to order cards for their catalogs. Special terminals allowed libraries to make changes to the cards, such as adding their own call numbers and locations, before they were printed. This allowed libraries to create cards for their catalogs more quickly than before. The cost of the cards was more than negligible, and they still had to be filed into catalogs, but libraries in that time felt that they were becoming more productive.
Beginning in 1980, I was working on the University of California's project to create an online union catalog for the nine campuses of the university. The first attempt to create a union catalog had been begun in the mid-1970's and was in the form of a book catalog. Libraries sent cards to a central office where they were assembled into pages for photo-reproduction. By the time the university was about half way through the alphabet, the early volumes were already out of date. But also during that time libraries had begun using a bibliographic utility that could deliver not only cards but a machine-readable version of the MARC record that represented the library's catalog entry, including the specific library holdings. By 1980 the university was producing a union catalog on microfiche that it delivered to each of the many main and branch libraries on the nine campuses. This union catalog used as input the machine-readable records that had been created as the side-product of card production and printing, and therefore contained data only from about 1976 forward. Each library continued receiving cards for its own catalog, but could now use the microfiche catalog to see the holdings of other campuses or of independent library units, such as law libraries, on its own campus.
When the online union catalog was unveiled in 1982, the larger libraries in the system were experiencing some difficulty keeping the card catalog up to date. One library reported that it was regularly running about 50,000 to 100,000 cards behind in its filing. This represented about three to six months of cataloging, and meant that a book could be cataloged and placed on the shelf many months before all of its cards would be entered in the card catalog. In short order, the union catalog, which received and loaded machine-readable records on tape on a weekly basis, became the most timely record of new library acquisitions.
Only a few years later the libraries had either purchased or developed their own local automation systems, and these had a public access interface. Each evening all of the new bibliographic records (and any updates to existing records) created that day were loaded into the catalog. The next day they were fully indexed and available through the user interface. Now the libraries faced a different problem: catalog records could be available before the books were on the shelf. Because it often took more than a day for the new books to reach their places on the shelves, users had to be instructed to check the "to be shelved" area for new books they couldn't find in the regular stacks.
At least one other interesting change took place around this time that was very noticeable to those of us working in the systems departments of the libraries and of the union catalog. When we first were receiving records through the bibliographic utilities, and while the libraries were using those utilities primarily to produce their card sets, a large library in our system contributed between 3,000 and 5,000 records to the union catalog each week. A majority of these were for new items or added holdings locations. Corrections to records were not unheard of, but the corrections had to be quite significant for the libraries to be willing to accept the cost. Every record updated incurred a fee through the utility as well as payment for replacement cards. Once received, those cards entered that filing backlog that I discussed earlier. When libraries had their own databases and record editing systems and were no longer producing cards, they could update or correct a bibliographic record without incurring a payment and without having to re-file cards. This meant that minor errors could be corrected more economically. In addition, library systems developed the ability to update multiple bibliographic records based on changes to authority records. Within ten years the same libraries that had once transmitted 5,000 records a week to the union catalog were regularly transmitting between 30,000 and 50,000 records a week, the majority of those corrections and additions to existing records.
These anecdotes reveal some ways that the use of library systems technology changed the overall efficiency of the cataloging function and also improved the quality of the catalog. How much change took place and what was the value of that change? To define that would require more than just these stories; it would require a set of data that quantified the work of the catalog department and its impact on the service provided by the library. Unfortunately, no one in my sphere gathered that data, although some of it may be discernable from library statistics.
The Association of Research Libraries (ARL) has gathered some standardized statistics from academic libraries since 1961. What the ARL statistics show over the period from 1963 to 20005 is that the cost of staff has sky-rocketed, about seven times what it was in 1968, the earliest date we have for this figure. But the cost for staff as part of the total library expenditures has dropped from about 55% to about 46%. The cost of "other expenditures," which is where library automation would be figured in this data, hovers at about 9% of the library's total expenditures through the 1960's, then begins a steady climb until it reaches about 25% of the total library expenditures in the year 2000, increasing 15-fold over that time in reported actual dollar amounts. Library expenditures themselves increase about 9-fold.
We can see that library expenditures have shifted from staff to … well, to something else. Unfortunately the ARL statistics did not have a separate category for the expenditures on computer hardware and software until 2004, and even now they do not distinguish between expenditures on the catalog and expenditures on other hardware and software that support library operations. What we can conclude is that library budgets would be even more stretched than they are today if we hadn't found a way to "do more with less." There is a good chance that technology played a major role in that change.
The gain during this time, however, is not just in shifting costs. The catalog today is more than the pre-automation catalog. It is available remotely, and at all hours. It often contains information on the availability of materials. It allows a variety of search capabilities that were not possible with the card catalog. It is more accurate, more up-to-date, and none of the entries are misfiled. This adds a level of complexity to determining the value returned from an investment in technology.
"Computer-based library system justification based on costs alone will continue to be difficult because machines not only replace manual systems but generally do more and different things."6
What we need is a way to measure those "more and different things." This may take us back to the concept of the social return on investment, the measurement of the value that the library provides to its community.
The ARL statistics have been primarily designed to count things (books, serial subscriptions, interlibrary loan transactions) and not to measure the productivity of library functions, like the creation of the catalog, much less the value aspects of accuracy. Staff productivity can be measured, at least in some areas of the library, and this would help inform decisions about the adoption of technology. The same is true for the more difficult measurements for aspects like quality and accuracy. Although any measurements for those less countable aspects of the library service may be somewhat artificial, they would still give us a basis on which to compare library functioning over time. We have at least two upcoming changes in technology that might benefit from just such an analysis.
The first is the implementation of RFID-based systems for the circulation of materials and to perform various inventory functions. These systems promise to greatly increase the speed with which one can check items in and out, as well as provide a reasonable security function (that is, no worse than any other system). A library with its materials tagged with RFID labels can perform shelf-reading and inventories through hand-held readers. On the cost side, at today's prices the cost of supplies for these systems is a greater than for systems based on barcode technology. What impact might such a system have on the overall library operating costs, and how would it affect the service the library provides? With forethought a library can gather data points that can be used to both predict and eventually track the changes in costs and benefits the library incurs with the implementation of this new technology. Because circulation is essentially a "countable" library activity, and one that has little qualitative value, those calculations should be relatively simple to make. The inventory function will need more artifice in its measurement, using staff time to perform the inventory without and with the new technology, and possibly a formula for the value of items found that had previously been considered missing.
The second is more difficult: the move toward mass digitization of library materials. There are project directors who have figured the costs of equipment and staff for the act of digitization.7 What hasn't been calculated yet, as far as I know, is the value of the benefits we might hope to gain from these projects. Much like our early move from card catalogs to the online catalog, this appears to be an obvious new technological direction although we can't seem to articulate what services will result, much less what they will be worth. Undoubtedly there will be some uses of these digitized materials that we cannot begin to anticipate today, but it would be highly desirable to have at least an hypothesis of the benefits we seek to obtain and some method of measuring those benefits. We can propose measures on access and delivery of materials that could be used in the future to compare the pre-digitization library to the post-digitization library.
I must admit that I shudder when I see an article carrying the bleak headline "Public Library Circulation Decreases, Expenditures Rise."8 Nowhere does the article even hint that circulation is no longer the best measure of the value a library delivers, nor that a number of those public libraries are more than making up for lower rates of circulation by delivering online services that greatly expand the amount of information available to its community. We do ourselves a disservice when we measure today's libraries against yesterday's services. With the vast digital resources that we make available to our users, the access to full text materials that are never "not on the shelf," the ability to search through huge bibliographic databases quickly and to download the results, there should be no question that libraries are worth every penny of the investment that goes into them.
Although new measures are being added to academic library statistics, such as DigiQUAL (Measuring Digital Library Service Quality), and MINES (Measuring the Impact of Networked Electronic Services) that were introduced recently by ARL,9 we still have this unfortunate tendency to overlook how the our investments in technology allow the institution to gain greater efficiencies. Yet the savings that we have achieved over the last twenty to thirty years may be making the difference between a library's ability to continue to provide excellent service in this fast-moving information society and going the way of the dinosaurs. Yes, it comes down to "evolve or die out," and if we aren't aware of our own amazing evolution, it's not surprising that our public fails to see it as well.
1 Investopedia. ROI – Return on Investment. Available: http://www.investopedia.com/terms/r/returnoninvestment.asp (June 5, 2006)
2 Jakob Nielsen. Jakob Nielsen's Alertbox January 7, 2003: Return on Investment for Usability. http://www.useit.com/alertbox/20030107.html (June 5, 2006)
3 Olsen, Sara. "Social Return on Investment: Standard Guidelines." Center for Responsible Business, Working Paper Series. Paper 8 (2003). Available: http://repositories.cdlib.org/crb/wps/8 (June 3, 2006)
4 Matthews, Joe. "The Value of Information in Library Catalogs." Information Outlook (July, 2000) 18-24.
5 This data is derived from a download of the median for all reporting libraries for the following data elements: Total Salaries and Wages, Total Library Expenditures, Total Other Expenditures.
6 Hugh C. Atkinson, Patricia F. Stenstrom. "Automation in Austerity." In: Austerity Management in Academic Libraries. Edited by John F. Harvey and Peter Spyers-Duran. (Metuchen, NJ, Scarecrow Press. 1984) p. 276.
7 Huges, Lorna. The Price of Digitization: New Cost Models for Cultural and Educational Institutions. National Initiative for a Networked Cultural Heritage. April, 2003. Available: http://www.ninch.org/forum/price.report.html (June 13, 2006)
8 Charity L. Cree, Mijung Yoon. "Public Library Circulation Decreases, Expenditures Rise." American Libraries. (November, 2005) 57-58
9 Association of Research Libraries. ARL New Measures Initiatives. Available: http://www.arl.org/stats/newmeas/index.html (June 13, 2006)
The copyright in this article is NOT held by the author. For copyright-related permissions, contact Elsevier Inc.