Return to Contents

CHAPTER FOUR

COMPONENTS OF LIBRARY EXPENDITURES

STAFFING

The long-term growth in number of staff in Research 1 libraries was steady between 1912 and the early 1960s, allowing for the inevitable fluctuations associated with two world wars  (fig. 4.1) [small | large] Taking the years since 1912 as a single period, we find that the total number of professional and supporting staff at Research 1 libraries has increased at an average annual rate of approximately 3.7 percent.

While all parts of this record are of historical interest, it is the patterns since World War II (and especially over the last three decades) that are most consequential. Distinct subperiods stand out. Following the war, there was a considerable rebuilding of staff (which lasted from 1944 to about 1949); once this was accomplished, staff size grew only modestly until about 1960. At that time the entire face of higher education began to change rapidly. The general expansion of colleges and universities had a dramatic effect on the size of the library staff as well as on library acquisitions. Between 1960 and 1970 the average size of the staff in the Research 1 libraries almost doubled (increasing by 94.7 percent), or an average of about 7 percent per year. While some of these increases in staffing were no doubt required by higher workloads associated with increased enrollments and the need to cope with an increasing volume of acquisitions (discussed later), another significant part of the explanation for this unprecedented growth in library staffing surely has to do with the equally unprecedented expansion of graduate education that occurred simultaneously. More branch and departmental libraries were established, and bibliographers, catalogers, and reference librarians were needed in larger numbers than ever before.

The growth in staff size then came to an abrupt halt. In the fifteen years between 1970 and 1985 the average size of the library staff at these same Research 1 universities increased by only 6.9 percent (less than 1/2 of 1 percent per year). Most recently, there has been a modest "recovery," in that the average size of the library staff grew by another 5.9 percent over the next six years.

More complete staffing data for all four of our composites (including data on total salaries paid, which are presented later) exist from 1963 on, and they show both the same high rates of increase in staff size during the 1960s and the same virtual halt to net additions during the 1970s  (fig. 4.2)[small | large][2] While the expansionary period of rapid increases in staff size lasted somewhat longer at the Research 2 universities than at the Research 1 universities, a sharp deceleration in the rate of new hiring followed by a period of essentially no change is evident in the plots for all four composites. In none of these sets of libraries, however, is there evidence of any absolute decline in the size of the staff during the 1970s. (The negative blips in a few individual years appear to be due to shifts in the numbers of temporary employees, associated with the birth and death of special projects.) The essential point is that there was very little net expansion in library staffs between the early 1970s and the mid-1980s.[3]

Another perspective on changes in staffing at Research 1 universities is obtained when we compare the growth in staff with changes in the size of collections. The long-term rates of increase in volumes held and volumes added per year have been roughly comparable to the long-term rate of increase in staff, with the average annual growth rate for volumes held (3.7 percent per year) exactly the same as the growth rate for personnel (3.7 percent per year) and the average annual growth rate for volumes added (3.1 percent per year) only modestly lower.[4] But beneath these long-term similarities are some sharply divergent trends.

The ratio of volumes held per staff member was reasonably steady during the pre-World War II years (fluctuating in the range of 10,000 to 13,000 volumes per staff member); it then fell sharply during World War II (to a low of about 8,000 volumes per staff member in 1948) before climbing back to roughly the 11,000 level in the early 1960s  (fig. 4.3a)[small | large]. During the decade of the 1960s the ratio of volumes held per staff member fell sharply and reached a level lower than that observed at any previous time (excepting only the World War II trough year). Since volumes held continued to increase after 1970, when staff size ceased to grow, there has been a steady and steep rise in the number of volumes held per staff member, which has continued to the present day.

An ever-larger collection is being managed by a staff that for the last 20 years has not increased at anything approaching a comparable rate. The result is that volumes held per staff member at the Research 1 universities is now at an all-time high of nearly 15,000 volumes.

The second measure of changes in collections---the annual number of volumes added gross, the measure of annual flow into the system as contrasted with the previous measure of the size of the stock---also declined in relation to staff size during the 1960s  (fig. 4.3b)[small | large]. The net additions to library staff were so great that they dominated even the substantial increase in the rate of acquisitions that also occurred during that decade (as well as, a fortiori, the increases in volumes held). However, while the ratio of volumes held per staff member then rose sharply, the ratio of volumes added per staff member continued to decline during the 1970s and 1980s (bottom panel of fig. 4.3b[small | large]). The reason for this decline is very different, however, from the reason for the decline during the 1960s: the retrenchment in higher education that began in the 1970s had an even stronger restraining effect on new acquisitions than it did on new staff.

In sum, then, library staff have faced contradictory trends in the early 1990s. They have had to manage a larger collection per staff member than ever before; at the same time, the number of new acquisitions per staff member has fallen back to the level of the mid-1950s.

BROAD SHARES OF LIBRARY EXPENDITURES

While absolute expenditures in all three categories have of course increased substantially over the last three decades, there have been some noteworthy shifts in relative shares. The basic pattern is most easily presented and described via snapshots taken at four different points in time: 1963, 1970, 1982, and 1991  (fig. 4.4)[small | large].

Expenditures on materials and binding at these 24 libraries have been a remarkably constant share of total library expenditures, moving from 33 percent to 35 percent, back down to 33 percent, and then back up to 35 percent. The noteworthy trends concern the other two components. Salaries have fallen steadily, from an average of 62 percent of total expenditures in 1963 to 52 percent in 1991. The remaining component, other operating expenditures, has risen just as steadily, climbing from an average of just 6 percent in 1963 to 14 percent in 1991.

These shifts in salary percentage may be more consistent with the trends in number of staff discussed in the preceding section than they first appear to be. What is surprising initially is the sharp decline in the salary share between 1963 and 1970 given the rapid increase in the number of staff between those two years. The explanation, we believe, is that the age distribution of the staff probably changed markedly during the 1960s, since most of the new additions to staff can be assumed to have been relatively young. Rapid increases in staffing are often accompanied by less rapid increases in payroll costs because of the growing fraction of the staff earning entry-level salaries. The post-1970 declines in the salary share of total expenditures require no special explanation, since they follow directly from the halt in recruitment described in the previous section.[5]

The increase in the share of total expenditures of the other operating expenditures category has been dramatic by any reckoning. This share has more than doubled, rising from 6 percent to 14 percent. By all accounts, increasing outlays related to computerization have been the driving force, and it is therefore not surprising that the largest jump in share occurred between 1970 and 1982, when the new technology was first introduced on a large scale.

Before commenting further on computerization, it is worth noting that the pattern just described has characterized all four of our library composites  (fig. 4.5)[small | large] with only slight variations among the sets of libraries. Shares of expenditures devoted to other operating expenditures started from a lower base in the Public 1 and Public 2 composites (5 and 4 percent, respectively) than in the private composites and then rose relatively rapidly in these two public composites. But there is little more to be said about differential rates of increase in the shares of any of these components. The point to emphasize is surely the commonality of the trends. (See also fig. 4.6)[small | large], which provides annual data for each of the four library composites.)

Pronounced similarities in the absolute levels of the shares across composites are also evident. In 1991 the share of total library expenditures devoted to expenditures on materials and binding ranged from a low of 33 percent in the Private 1 composite to a high of 36 percent in the Public 1 composite. The salary share in 1991 ranged from a low of 50 percent in the Private 2 composite to a high of 53 percent in the Public 2 composite (with values of 52 percent in both Private 1 and Private 2). And the share devoted to other operating expenditures in 1991 ranged from lows of 12 percent in Public 1 and Public 2 to a high of 15 percent in Private 1 and Private 2.

OPERATING EXPENDITURES AND COMPUTERIZATION

At many institutions the first of the functions to be automated was circulation, which was seen as an obvious candidate for such treatment. At many institutions barcode labels that are optically scanned at the moment the book is charged out have replaced cards, which in the past had to be removed manually from the volume and filed. The advantages of storing the title, call number, and author's name electronically are obvious: any of the elements can be retrieved and a patron need know only one element of the three in order to request information about the volume.

The automation of the cataloging function has been of even greater importance; indeed, in Warren J. Haas's words, "[c]ataloguing is what turns an accumulation of material into a library collection,"[6] and the technological advances of the past few decades have afforded a degree of standardization within the entire system nationally that previously would have been impossible. In the past a professional cataloger at a particular institution would either write his or her own catalog copy or make use of cards provided by the Library of Congress, which sometimes required revision resulting from a different classification scheme or minor differences between the volume actually in the possession of the cataloger engaged in writing copy and the volume cataloged by the Library of Congress. [7] Subsequently, Library of Congress copy was made available online from one of the national vendors, in particular OCLC (originally the Ohio College Library Center, now the Online Computer Library Center) and RLIN (the Research Library Group's Research Libraries Information Network). Individual member libraries also contribute records written by their professional catalogers to both the RLIN and OCLC databases.

The advantages of this approach are obvious: copy made available online can readily be reformatted to fit specifications peculiar to a particular institution. Moreover, because member institutions contribute copy there are additional sources of records in the absence of Library of Congress copy, which obviates the need for original copy to be written at each individual institution and offers the promise of greater uniformity in the content of the records. In the past the Research Libraries Group charged member institutions that used the RLIN database a per-transaction fee. Members may now purchase transactions in blocks (500,000 searches annually, for example) and are given rebates for each record contributed to the database.

The acquisitions function has also benefitted from automation. The Research Libraries Group, for example, provides a service that enables a member library to search the database for information about items it may wish to purchase, and order forms may then be generated directly from the database. Similarly, invoices in machine-readable form are often included with shipments of books. The information that the invoices contain is integrated into computer files and affords a means of achieving various kinds of control over the acquisitions budget generally and the performance of vendors.

These technological developments mean that while the basic library functions continue to be performed, they are now performed in very different ways. And the costs of performing them are now allocated somewhat differently among categories of expenditure than they were in the past. In effect, trade-offs have been made between staffing and automation---which is not to say that this has been a conscious, carefully articulated process. As Kendon Stubbs has put it, in commenting on an earlier draft of this manuscript:

It is true that other operating expenditures have risen faster than total expenditures and staffing somewhat less. But I doubt that any library director would say that over the past 15 or 20 years there has been a deliberate shifting of money from staff to automation and other "other operating" expenditures. In fact, as you note, the absolute size of ARL staffs has never declined; it has just grown more slowly in recent years. Thus, in the various places in this study where you suggest a conscious trade-off between staffing and automation, I would be more inclined to suggest something less planned. ARL libraries tend to be conservative and to hold onto staff, often for traditional functions even when those functions are no longer cost-efficient (though in the very recent past, under the impact of the recession, there is evidence that the libraries are being more hard-nosed about relinquishing staff in traditional but low-impact functions).

It may be closer to the truth to say that by the mid- seventies new staffing positions were harder to come by than they had been in the sixties, so that increases in staffing (and staffing expenditures) slowed; while the then relatively small amount devoted to other operating expenditures was allowed to grow at its own pace, chiefly driven by automation. I don't know if this formulation comes out sounding really different from yours; but it does propose that the flow of money from staff to other operating was more fortuitous than planned up to very recent years. One piece of evidence for the unplanned nature of the historical trend is that during the late seventies and up to the past few years it was reported as axiomatic in library literature that automation does not save staff. If you were a library director and were requesting funding for an online cataloging system, you had to sell the concept to your administration while at the same time telling them that you could not give up any cataloging staff after you were automated. This was an unrealistic sales job, even if the library community had convinced themselves that it made sense; and university administrations may have reacted by putting reins on new staffing, while hoping that automation would stabilize library costs.[8]

We conclude this part of the discussion by noting that it is difficult to determine with precision how the effective functioning of the library has been affected by the redistribution of shares of library expenditures from staffing to automation. The reason is that the automation of circulation, cataloging, and acquisitions has changed the nature of these functions. Our strong impression is that the quality of these services has been enhanced, in some instances quite appreciably, but this is hard to prove definitively.

EXPENDITURES WITHIN THE MATERIALS CATEGORY: SERIALS

In both the Research 1 and Research 2 composites the percentage of the materials budget devoted to serials increased rapidly during the 1970s and peaked about 1981.[11] After a period of decline in the early 1980s, the serials share began to rise again beginning in 1986 (fig. 4.7)[small | large]. The main difference between the Research 1 and Research 2 composites is the absolute share of the materials budget devoted to serials. Throughout this entire period the Research 2 libraries spent approximately 10 percent more of their materials budget on serials than the Research 1 libraries. This difference apparently results from the fact that the materials budgets of the Research 2 libraries are generally smaller. At the smaller institutions librarians may feel that first priority has to go to purchasing a reasonably comprehensive set of serials.

The somewhat erratic path of this time series is due to varied rates of increase in expenditures for serials. Serials expenditures have increased rapidly for the entire period since 1976, but three subperiods can be distinguished---1976-81, 1981-86, and 1986- 91---with the middle period one of somewhat more moderate increase and the last the period of most rapid increase  (fig. 4.8)[small | large].Between 1986 and 1990 the All 24 composite increased at an average annual rate of increase more than 11 percent.

Trends in the prices of serials are discussed at some length in Chapter 6. We can anticipate that discussion by noting that the price increases have been the driving force in increases in the serials share of the materials budget. Evidence for this assertion is the fact that larger and larger expenditures for serials have not led to a comparable increase in the number of serials acquired  (fig. 4.9) [small | large] In fact, between 1986 and 1990 the number of serials received at Research 1 libraries actually decreased by 6 percent.[12] During this same period, nominal expenditures increased by 73 percent.

As we will discuss in Chapter 6, price increases for journals vary significantly by field, and the experiences of individual universities illustrate the effects of these increases on serials budgets in recent years. For two individual universities for which we were able to collect data (a Public 1 university and a Private 2 university), the percentage of the serials budget expended for science serials has increased steadily since the mid-1980s  (table 4.1) [small | large]. At the Private 2 university we calculated the science serials expenditures as a percentage of all serials expenditures for humanities, social sciences, and science departments. (General serials expenditures were not included in the total.) This percentage has increased from 58.9 in 1983-84 to 62.9 in 1990-91. Although data on the number of serials were not available, we have been told that the total number of serials stayed relatively constant, with the few serials that were added concentrated in the humanities and social sciences rather than in the sciences.

At the Public 1 university, expenditures for science serials as a percentage of the total were remarkably comparable to those at the Private 2 university. In this instance, expenditures for science serials increased from 59.1 percent of the total (including humanities, social sciences, sciences, nonscience libraries, and area studies) in 1985-86 to 64.2 percent in 1990-91. Again, data on the number of serials are unavailable for those years, but we can get an idea of the magnitude of these figures with data from 1991-92. In this year 1,947 science serials constituted only 28.7 percent of the total number of serials in the same areas included in our total expenditures---that is, expenditures for science serials constituted approximately 65 percent of the serials budget and provided approximately 29 percent of the total number of serials. [13] Conversations with librarians at other universities suggest that percentages such as these are not uncommon.

The rapid increases in serials expenditures documented earlier and the substantial redeployment of materials expenditures toward serials and away from monographs may now be more fully understood as responses to external forces that have had such pronounced effects on the functioning of academic libraries. Indeed, it would not be an exaggeration to say that of the various factors in the constellation affecting university libraries in recent years, the rapidly rising prices of periodicals have in many respects been the most important. They explain the de facto encumbering of the materials budget, and they surely go a long way toward explaining the widening gap between the numbers of volumes added gross and of book titles published, since the redeployment of the materials budget toward serials has constrained libraries' ability to purchase monographs. [14] As we have seen, there has been little increase in the number of serials acquired annually, and these developments can therefore be understood as retarding the growth of the collections. It is not surprising, then, that in writing of the situation, authors have invoked such metaphors as "the library doomsday machine" and "the journal that ate the library."[15]

There are a number of possible ways of looking at the inability of libraries to accommodate fully the changes in the serials and book industries. Both internal and external factors are involved. Emphasis could be placed, on the one hand, on the inadequacy of the materials and binding budget. Since materials and binding expenditures increased at a lower rate than total educational and general expenditures throughout the 1970s and 1980s, one might argue that the level of institutional support for acquisitions was inordinately low and that an appropriate response in light of the increases in both title output (books and serials) and the prices of library materials would have been to increase the percentage of the educational and general budget expended on the acquisition of library materials.

On the other hand, there are always many competing claims on university resources, and each such claim, no matter how important, has to be evaluated in relation to the others. On this basis one could argue that the relevant external forces, no matter how pressing, simply did not justify the extraordinary redeployment of institutional resources that would have been necessary had individual universities attempted more comprehensive coverage. The issues are not unlike those involved in evaluating the continued viability of a need-blind admissions policy, for example. Such a policy remains an exceedingly important objective at many institutions and is considered vitally important to institutional health. Nonetheless, there are other critically important objectives---compensating faculty members adequately, supporting faculty research activity adequately---that also require attention.

In that sense, the forces contributing to the widening gap between rates of increase in volumes added and rates of increase in title output can be said to have been principally external rather than internal, in that the external demands were far too great for individual institutions to undertake the kind of response necessary to meet them more fully. Those demands, in short, were out of phase with institutions' ability to respond.

Endnotes

[1] Reliable data on shares of total expenditures date back only to 1963. However, cruder data on salary budgets, seen in relation to expenditures on materials and binding, go back to 1912 for at least some libraries. Examination of the relevant ratios suggests strongly that salaries have long been the largest single category of expenditure. (At Berkeley, for example, salaries were approximately 1.5 times expenditures on materials and binding during the decades before World War II; this same ratio was approximately 2.1 in 1963 and 2.2 in 1991.)

In making this statement and in presenting other expenditure ratios in this chapter, we follow---of necessity---the traditional practice of taking total direct expenditures by libraries as the base for all such calculations. This method is seriously misleading in that no weight is given to the costs of space, which are very considerable. (In Part 2, we speculate briefly on the importance of taking the costs of space into account if wise decisions are to be made concerning new publishing forms made possible by electronic technology.) The neglect of space costs is but one example of a more generic problem with university accounting---namely, the tendency either to understate or to ignore altogether capital costs of many kinds. See Gordon Winston, "Why Are Capital Costs Ignored by Colleges and Universities and What Are the Prospects for Change?" Williams Project on the Economics of Higher Education, DP-14, July 1991.

[2] The annual data presented in figure 4.2[small | large] are calculated as three-year moving averages of the year-to-year change in the size of the total staff.

[3] This virtual halt in the growth of professional and nonprofessional staff was accompanied by an increase in the number of student assistants. In all but the Public 2 composite the ratio of student assistants to total staff increased modestly throughout the 1970s. At the Private 1 composite, for example, this ratio increased from 0.168 to 0.196, an increase of approximately 17 percent. However, not even the larger number of student assistants, when expressed at Full-Time Equivalent (FTE) employees, alters the basic proposition about slow growth of staff during the 1970s. The absolute number of student assistants added during the entire decade ranged from only 10 to 30. The changing mix of staff itself reflects changes in the way that libraries operate.

[4] See Chapter 2, figures 2.1[small | large] and 2.2[small | large]. The average rate of increase in volumes held (which is a stock) will normally be greater than the average rate of increase in volumes added (which is a flow). The total number of volumes held would still increase somewhat (albeit at a declining percentage rate) even if the growth rate for volumes added were zero, since a zero growth rate for volumes added would be consistent with a constant annual increase in the absolute number of volumes added. This is, in fact, a reasonably accurate description of what has happened to collections over the last two decades.

[5] These salary figures include the salaries of student assistants.

[6] See Haas's preface to The National Coordinated Cataloging Program: An Assessment of the Pilot Project (Washington, D.C.: Council on Library Resources, 1990), v. It should be noted that under Mr. Haas's leadership the Council on Library Resources played a continuing role in supporting shared cataloging, automated catalogs, and linkage of bibliographic systems.

[7] Since the establishment of the Cataloging in Publication program, moreover, an effort has been made by the Library of Congress to write catalog copy for books before their publication, so that it could be printed in the published books. Because various elements of catalog records written prospectively do not always conform precisely to the books in their final published form, the records cannot in all instances be accepted without revision.

[8] Letter from Mr. Stubbs, associate librarian for public services, Alderman Library, University of Virginia, to Anthony C. Cummings, January 29, 1991.

[9] We are reminded of a famous controversy in economics in the 1940s in which Fritz Machlup and Richard Lester debated the extent to which business decisions were in fact based on comparisons of marginal costs with marginal revenues. Machlup used the analogy of the driver of an automobile making a decision whether or not to overtake a truck. The driver (we all agree) does not consciously make the complex calculations needed to determine the rate at which he must accelerate under varying conditions; still, he drives as if he had made the calculations. See Fritz Machlup, "Marginal Analysis and Empirical Research," American Economic Review 36 (1946): 534-535.

[10] For the rest of this discussion we will focus on expenditures for materials only (as opposed to materials and binding). We use the terms "materials" and "acquisitions" interchangeably.

[11] Because data were either not available or unreliable, Cornell University and the University of Florida are not included in their relevant composites for the following discussion of serials and serial expenditures.

[12] The data on the number of serials received that are presented in figure 4.8[small | large] and discussed here include both serials purchased and serials not purchased. Data for serials received have been collected in these two categories since 1986, and a better comparison with serials expenditures would of course be with serials purchased only. Unfortunately, we included only the total category of serials received when we created the original spreadsheets.

[13] Moreover, neither of the two universities mentioned has an engineering department, which would likely have made the science expenditures an even higher percentage of the total.

[14] Universities have attempted to cope with this problem in a variety of ways. At 1 of the 24 institutions studied here, the realization that there was a widening gap between rates of increase in acquisitions and rates of increase in international book title output resulted in a decision to relate levels of funding for acquisitions to increases in the number of titles published in the United Kingdom and the United States, though the decision entailed a substantial increase in the acquisitions budget.

Other universities have attempted resource-sharing initiatives. James Madison University, the University of Virginia, and Virginia Polytechnic Institute and State University, for example, agreed to provide one another with copies of articles from journals not in their own collections; the copies are sent by telefacsimile within 24 hours, and publishers are compensated appropriately. What these initiatives and other, similar ones have in common is that they apply existing services and technologies---interlibrary loan, photocopying, telefacsimile---to the problem of access to materials held elsewhere.

It is important to remember that, despite initiatives such as these, the primary way of dealing with the serials crisis has been to redistribute the acquisitions budget towards serials and away from monographs. (Dorothy Milne and Bill Tiffany, "A Survey of the Cost- Effectiveness of Serials: A Cost-Per-Use Method and Its Results," The Serials Librarian 19 (1991): 137-149, and "James Madison University, Carrier Library, Documents Express Program." We are grateful to Dennis E. Robison, university librarian at James Madison, for providing a copy of this last item.)

[15] Ann Okerson and Kendon Stubbs, "The Library `Doomsday Machine,'" Publishers Weekly 238 (February 8, 1991): 36-37; Herbert S. White, "The Journal That Ate the Library," Library Journal 113 (May 15, 1988): 62-63.