Robert M. Hayes
Graduate School of Library & Information Science
University of California at Los Angeles
Los Angeles, CA 90024, USA
E-mail: rhayes@mvs.oac.ucla.edu
The context for this paper is a model for assessment of library staffing and related costing patterns as functions of workloads on library services and operations. The model, called LSM -- The Library Staffing Model, estimates staff requirements and costs using what are called "workload factors". They relate workloads, measured by units of work appropriate to each type of service or operation, to staffing required to handle those workloads. The resulting staff estimates are by categories of staff (e.g., professional, support, and hourly) and by function; they are measured in full-time equivalents (FTE).
The default values used in LSM were first published twenty years ago, but they remain essentially the same in its implementation today. During that time, though, automation has become a standard part of library operations and services. So repeatedly, as the LSM has been presented to librarians, the question has been asked, "Hasn't automation affected the efficiency of staff? Why haven't the workload factors changed?" On the surface, that question is reasonable. In fact, it is being repeatedly asked in commerce and industry where expectations of improved clerical productivity as a result of automation have almost never been met.
This paper will briefly summarize the general nature of LSM and then will discuss the effects of automation upon staffing and more general aspects of library operations and services using the framework of the LSM components:
• Effects on workload factors
• Effects on staff and associated direct salaries
• Effects on direct expenses for information materials and operations
• Effects on overhead (especially for equipment amortization and space)
• Effects on general administration
2. A BRIEF SUMMARY OF THE LIBRARY STAFFING MODEL (LSM)
LSM is based on six matrices, three of clients and services for them and three of materials and technical processes for them. In each case, the first matrix contains data for determining workloads involved; the second contains data for the extent to which workloads use specific services or processes; and the third contains workload factors as means for estimating required staff and costs.
2.1. The Measurement of Workloads
In the model, the workloads are considered in two major categories: those related to services to the populations served and those related to internal (technical) processing of the materials acquired. In each case, the model separates the measurement of workload into two matrices, the first measuring the size of the populations involved and the frequency of use and the second measuring the distribution of uses across the set of services provided.
The value of this approach in developing the model is that it separates the tasks of measuring the populations from that of measuring the uses of services. In that way, each can be investigated, without worrying about the other.
The value of this approach in applying the model is that, given the matrices measuring using populations and their use of services, the workload for each of the services can be calculated by simply multiplying the two matrices. In the same vein, given the matrices measuring materials acquired and their use of processes, the workload for each of the processes can be calculated by simply multiplying those two matrices.
It is of interest to note that the two values, percentage and frequency of use, are a simplistic way of representing the typical highly skewed distribution of use of an information resource -- what is called a "J-shaped" curve. It would be of value eventually to represent it by more fundamental parameters, since that could give insight into the effects of the J-shaped distribution.
Obtaining and verifying data for this matrix should be relatively simple. Of course, it requires that the set of populations be identified, but that surely derives directly from the identified mission of the library. The first three columns of data should be readily derivable from reports and internal records of operations in either the institution or the library itself.
• Distribution of Use of Services. The second matrix has the same rows as the first, but the columns now relate to the set of services provided by the library. In LSM, those services normally would include such services as circulation and inhouse use of materials, photocopying use, multi-media-av services, ready and extended reference, analysis and consultation, instruction, database access, access elsewhere (i.e., ILL and document delivery).
The value in each column, for a given population, measures the percentage of uses of the library that will entail the service related to that column. It is these values that provide the means for LSM to derive the workloads on each of these services. Getting the values for this second matrix is clearly more difficult than for the first, but it is still quite feasible to do so. Values can be estimated from knowledge of the patterns of use of a given population and then be tested by observation. Other values can be estimated from statistics on services which may identify the category of each user. Some values may be obtained by sampling of users during a period of time, by interviews or by questionnaires.
• Types of Materials. The types of materials currently included in LSM are as follows: monographs, monograph series, periodicals, microforms, special collections (such as rare books, manuscripts, photographs, etc.), documents (such as government documents or vertical files), media (including both audio-visual and computer media), and special projects (such as and indexing and abstracting services, backlog processing, etc.).
The matrices for types of materials consist of rows, one for each type of material acquired. In the first matrix, the first column contains a measure of overall holdings of the type of material; the next three columns relate to various levels or processing or status for the materials currently held; the fourth column measures the expected additions to the collection; the final three columns measure the transition of materials between levels of processing or status.
• Distribution of Use of Processes. The second matrix has the same rows as the first, but the columns now relate to the set of internal (technical) processes carried out in the library. In LSM, those processes include acquisition of materials, cataloging at various levels of detail (i.e., accessioning, full cataloging, and producing finding aids), preservation and conservation, and various types of physical handling, as listing in the following table.
The value in each column, for a given category of materials, measures the percentage of materials being processed in the library that will entail the specific service related to that column. The values for this second matrix are surely derivable from operational records in operating departments.
Workload factors are the basic means for estimating staff and costs. The discussion will first set a context by discussing various approaches to costing so that the use of workload factors is seen in context. It will then look at the workload factors themselves. It will conclude by considering other issues related to staff and costs estimation, such as overhead.
It is important to recognize the differences among these approaches, since they result in dramatically different estimates of cost that are difficult to reconcile. The most accurate and complete is a true cost accounting based on records at the time costs are incurred and properly allocated. The most detailed will be the time and motion study, but it will usually account for only the most specific costs, not dealing with benefits or with allocable costs since it measures only the actual time in performance of a defined set of tasks. At the other extreme, the total cost approach provides no detail and no means for analysis of functions; it is the least accurate of the methods and may grossly mis-estimate the costs, both under and over, in ways that make it impossible to calibrate. The ad hoc cost accounting approach used in LSM is an effort to establish a standard means for dealing with quoted costs that will include all components in a framework that permits analysis.
• Estimates of Direct Labor: The Use of Workload Factors. The LSM uses a matrix of "workload factors" as the means for estimation of the staff required to handle a defined workload, measured in appropriate "units of work". For each of a set of library services and processes, such units of work have been defined. The workload factors then are expressed as percentages of FTE (yearly full-time equivalent) staff for the performance of 1000 (K) units of work for each library function and subsidiary process.
The means used for expressing workload factors (FTE per thousand transactions) may seem unfamiliar. An alternative means for expressing them could have been "minutes per transaction". To translate from one to the other, note that a working year is almost exactly 100,000 minutes (i.e., 40 hours per week for 42 working weeks is 100,800 minutes), so 1 minute per transaction would be .01 FTE per 1000 transactions.
The LSM incorporates default values for the workload factors, but they can be modified to reflect specific library experience. The default values currently used in LSM are, with very few exceptions, identical with those first published in 1974 (Hayes and Becker, 1974)
The default workload factors in LSM are consciously and deliberately used with minimal precision -- typically on the order of 10%. This reflects the fact that there is great variability among institutions and, even within a single one, differences in the qualitative character of the workload. It would be spurious to imply high levels of precision simply not warranted. As a result, there is no reason to anticipate that use of these default values in a given library will match actual data any closer than the precision implies -- i.e., within say 10%. It is for that reason that the LSM provides means to change the defaults.
However, the default values for the workload factors are reasonable starting points for assessment. They were developed by an iterative process that involved hundreds of libraries, of all kinds and sizes, over a period of decades. The uniform base of the process was the fundamental cost accounting structure into which the observed data were fit. Estimates at a given point in time were matched with actual and/or reported costs for a visited library and operations within it as of that time. If estimates matched library data to within about 10%, they were regarded as further confirmed.
But if the estimated costs differed by substantially more than 10% from the reported ones, they were carefully examined for possible reasons for the differences. Were they caused by flaws in the workload factors; if so should there be changes in them? Did they reflect differences in functions and processes; if so should there be additions to them? Were there differences in efficiencies; if so, how should the efficiencies be treated?
Beyond the use of actual data from visited libraries, other means were used to determine default values. A number of time and motion studies were conducted to determine production rates on typical generic library activities. The rich array of published data were reviewed and reduced to a consistent accounting structure. Analogies were drawn among various kinds of operations, both within libraries and between them and industrial and commercial counterparts. To put these various sources of data into a common framework, rule of thumb ratios were used for conversion to fully burdened estimates. (Specifically, the ratio of "Direct FTE" to "Time and Motion FTE" was taken at 1.5/1 and the ratio of "Total FTE" to "Direct FTE" was taken at 1.5/1).
• Overhead. The next major component of the LSM cost accounting model is "overhead". In general, overhead or, as it is frequently called, "indirect" or "burden", includes those costs that cannot be directly attributed to "productive work". That in no way means that such costs are not necessary or significant. It simply means that it would be difficult or irrational to attempt to associate them directly with productive work. Supervision, for example, is clearly necessary, but it in no way in itself produces catalog entries, binds volumes, films pages, provides reference services, or does the actual work whatever it may be. Holidays, vacation, and sick leave are necessary components of a compensation package, but usually no productive work is accomplished during such time.
The purpose of "overhead" or "indirect costs" is to account for these kinds of costs. A typical rule for doing so is "proportional to direct labor costs", and that is the rule used in LSM. There are two alternative means for presentation of the overhead accounts. One, which treats "total salary" as the foundation, is typically used in "not-for-profit" institutions; the other, which treats "direct salary" as the foundation, is typically used in commercial contexts. Whichever may be used, the total costs (leaving aside allocation of profit or amortization of capital costs) must be the same, if efficiencies are the same. And, with all due respect to the much vaunted efficiency of private enterprise, there is nothing inherent in the tasks here that implies greater efficiency as a result of different means in accounting for costs. For a variety of reasons, though, the commercial accounting model is the one used in the LSM.
I turn now to the primary concern of this paper: the effects of automation upon the structure and values incorporated into LSM.
3.1. Effects on Workload Factors, Staff, and Associated Direct Salaries
The evidence from application of the LSM to data from hundreds of libraries over the past thirty years is that automation does not fundamentally change the efficiency of human operations. Persons still keyboard data at about the same rate; they make errors at about the same rate; they make decisions at about rate; they handle physical objects at about the same rate. Since these are the things that the workload factors measure, there is no reason to expect them to change.
What does change is the mix of tasks performed. That is best illustrated by cataloging, in which the mix of original cataloging and copy cataloging has dramatically changed in the past twenty years. The effect of bibliographic utilities has been profound and has led to dramatic reductions in the staffing of technical services, with shifting of staffing from them to user services.
One effect of new technology may lead to change, not so much in the workload factors themselves as between them and the overall operation. It is the potential of "telecommuting", in which work may be done remotely thus reducing time lost in transportation and increasing the morale of staff.
Even more important, though, is the effect on services and on the users. It permits the library to provide services that would otherwise be inconceivable, vastly increasing the value to the users, and improving their productivity. The result has been increases of staff requirements for providing information services and for educating users in the opportunities provided by the new information technology. It is this that makes the new technology worthwhile.
Perhaps the most important effect of new technologies is the increasing need for reference librarians to serve as consultants, advisors, and teachers in the use of them. In the both academic and public libraries, that implies expanded bibliographic instruction to include new technologies. In academic libraries it has involved establishment of centers for assistance to and guidance of faculty in development of curricula and new methods of classroom instruction based on new technologies.
In this respect, it needs to be recognized that the use of new information technologies in classroom instruction is by no means simple. It is likely to require a complete re-thinking of the structure, contents, and means for presentation. While faculty in the future will be increasingly knowledgeable in use of the technologies for their own research and writing, that by no means implies that they will be expert in creating instructional packages. In fact, the time and effort required to do so will certainly be substantially greater than is involved in writing a new textbook, for example. Furthermore, the development of new instructional packages requires the availability of the most advanced equipment which, because of its costs, needs to be shared among faculty, so the library's center for faculty instructional development serves as the means for making them available economically.
3.2. Effects on Direct Expenses for Information Materials and Operations
The acquisition of new information technology materials is just beginning, but already CD-ROM acquisitions are part of the budget of every academic library and many of them are acquiring magnetic tape files to load into local systems for access through campus online public access catalogs. According to Predicast data, this form of publication is increasing at a compounded exponential rate of 50% per annum -- more then a doubling every two years. As related new information technologies, multi-media systems in particular, have become widely available they have added further pressures, and within this decade they will probably consume up to about 10% of the library's acquisition budget.
The effects of the Internet and World-Wide Web, with the ready access to data and document delivery services that they provide, while now just beginning to be felt, will be profound. The problem, though, is that it is by no means clear what they will be nor how soon they will occur, since they depend upon the pace at which publishers move to these means for distribution. The likelihood is that scientific and technical journals, especially those produced by commercial publishers, will move rather rapidly to publishing of journal articles, rather than issues, through online electronic delivery. If so, the delivery costs may well be borne by the user rather than the library, but the problem will be how to serve the needs of those (such as students) without the financial resources. For more general, popular journals and for monographs, the pace is likely to be much slower.
3.3. Effects on Overhead (Indirect Salaries, Equipment Amortization, and Space)
Will new information technology affect indirect salaries? That is a question of some current debate. Some have argued that information technology can significantly reduce the need for middle management and for supervision in general. Their view is that distributed computing will permit the elimination of layers of management, replacing them by inter-communication through shared data files.
However, there are even more persuasive arguments against that view. Middle management is needed to assure that data are valid. They are needed because "span of control" is still subject to the reality of what persons can do. They are needed as means to communicate between strategic, tactical, and operational levels of management. They are needed as the training ground for top levels of management.
With respect to the equipment and software associated with the new information technologies, the direct cost implications are large and steadily increasing. While the costs of computers themselves continue to decrease in principle, (i.e., in terms of the cost for the same performance) they are not decreasing in fact, because the systems require increasing resources. The more important effect, though, is the number of devices that are needed to serve the needs of users of the new information technologies. In particular, the number of wired carrels in libraries of every kind is dramatically increasing, and even the traditional library table is being set up for power and telephone jacks that will allow users to bring in their own laptops. The growth of on-demand access to full-text requires that there be printers available to meet the needs in output of the accessed materials. For many applications, there need to be scanners for input.
All of this has meant that increasing amounts of space in libraries are being committed to the equipment associated with the new information technologies. If proper amortization for capital costs -- in purchase of hardware and software, in conversion of data files, in installation of facilities, in the renovation of space -- were included as part of overhead, the effects would be substantial. This is more than simply an accounting issue, because the nature of new information technologies is that they require replacement at intervals on the order of three to five years. First, even though electronic, these devices deteriorate over time and must be repaired or replaced; second, the pace of change is so great that they become obsolescent if not obsolete and must be replaced to assure that the needs of users continue to be met.
3.4. Effects on General Administration
For at least the past two decades, automation and the effective utilization of new information technology has been a major focus of strategic management for academic libraries. It has required the attention of library directors in order to obtain the funds required, hire the staff needed to implement, control the processes in acquisition of hardware and software, allocate the space and operating resources needed to house them, participate in cooperative efforts to minimize costs by sharing bibliographic data, deal with the needs of staff in adapting to the changes in operations, inform the academic community of users and work with them to assure their needs are met. Taken together, these tasks may well have consumed as much as 25% of time for academic library directors.
Addition of staff -- systems staff and operating staff -- has been required to make the necessary technical decisions and to implement and manage the results. The percentage of G&A staff involved in such efforts appears to be about 10% of total general management for a large library and, for smaller libraries, an even greater proportion (because they lack the economies of scale). Frequently their work requires hiring consultants and short-term project staff to support the internal staff.
It is evident that new information technology will continue to require at least at comparable levels of commitment at the highest levels of library management. The costs for doing so will probably be in the range of 10% to 15% of the G&A budget, or about 1% of total library budget.
4. SUMMARY
In summary, the LSM, The Library Staffing Model, has been presented as a frame of reference for assessment of the effects of the new technologies on library staffing and associated costs. The bottom line with respect to staffing is that the new technologies imply the shift of staff from internal operations into reader services, especially in consultation and instruction, and into information systems support.
With respect to costs, despite the wishes of administrators
that electronic delivery of information will imply a reduction of them,
the facts are that there will be increasing costs for equipment and related
software as well as for the information itself. The manufacturers of new
technology equipment and software are there to make the dollars, even billions
of them, not to give to the public good. And the publishers are not about
to reduce the costs for the information they distribute, whatever may be
the means for doing so.
REFERENCE
Hayes, Robert M. & Becker, Joseph. (1974). Handbook
of Data Processing for Libraries. 2nd Edition. New York: John Wiley &
Sons, Inc. Chapter 4, page 118, "Cost Accounting in Libraries."