Cataloging Staff Costs Revisited.

Staff costs for cataloging have declined at Iowa State University Library. This is demonstrated by data from a longitudinal time and cost study begun in 1987. We discuss the national developments, technological advancements, and reengineering efforts that have supported greater cataloging effectiveness and quality. We use the ISU findings as an example of a nationwide phenomenon resulting from the remarkable ability of catalogers to share work through national bibliographic utilities.

In 1987 the Technical Services Division of the Iowa State University (ISU) Library initiated a time and cost study to investigate the impact of automation on services and products. This study, now in its thirteenth year, has resulted in a number of reports in the literature. The earliest of these provided an overview of cataloging costs (Morris 1992) and a comparison of costs for serials and monographs cataloging (Morris and Osmus 1992). Since then, refinements in the analysis of tasks and costs (and especially in the application of staff overhead) have made more sophisticated and focused reporting possible. At the same time, however, these refinements preclude easy comparison of the earliest three years of the study to the years following.

In the present article, then, we report changes in cataloging costs and productivity since 1990 and discuss the factors contributing to these changes. Morris, Rebarcak, and Rowley (1996) previously noted some of the trends presented here. Morris and Wool (1999) presented a brief discussion of these trends in relation to the value of cataloging.

Literature Review

The literature on cost studies for technical services operations is extensive--as is evident in bibliographies from Dougherty and Leonard (1970) and Tavenner (1988)--but for the most part it is fragmentary, limited in scope, and short on detail. In much of this literature, researchers either estimate in-house operating costs for comparison with prices for vendor-supplied services or offer models for cost-benefit analysis. Of the rest, Lancaster (1977, 265) provides this assessment:

A number of studies on technical processing costs have already been published.... While several ... appear to be very thorough and complete, cost analyses of this type generally have two basic limitations: (a) although many data are presented, it is not always clear how these data were derived, and it is thus impossible for a second investigator to duplicate the methodology to obtain truly comparable data for a second institution or group of institutions, and (b) directly related to the first point, there are no generally accepted standards for what should be measured in these cost studies and for how the costs should be derived and presented.

This statement is just as trenchant with regard to the subsequent literature, highlights of which include Getz and Phelps (1984); Valentine and McDonald (1986); Leung (1987); Oldfield (1987); and Fiegen, Heitshu, and Miller (1990). Harris (1989) offers an interesting survey of publications on cataloging costs, along with an estimate that cataloging costs between 1876 and 1986 rose 4200%, much faster than general inflation but slower than librarian salaries.

Relatively few examinations of cataloging costs have appeared since Morris (1992). In the most extensive report, Jenda (1992) presents a workflow analysis and time/cost study made to support a decision at the University of Botswana on continuing the library's subscription to Library of Congress (LC) catalog cards. In this study, times for cataloging tasks were measured in an experimental setting. Byrd and Sorury (1993) document a significant time/cost analysis of authority work at Indiana University. EI-Sherhini (1995), in an evaluation of outsourcing the cataloging of Slavic-alphabet materials at Ohio State University, includes a brief cost analysis of doing the work in-house. Rider and Hamilton (1996) report tests of the OCLC Online Computer Library Center, Inc. PromptCat service at Michigan State and Ohio State universities, with a cost/benefit analysis based on estimates of staff time and costs as well as other data.

Time and Cost Method

A detailed description of the method employed in this study appears in Morris (1992). A more concise description, reflecting the changes made in 1990, can be found in Rebarcak and Morris (1996). Highlights are recapitulated and more recent developments in the method are presented here.

Data Collection

Five times each year, every technical services staff member, including hourly student employees, tracks all time worked for an entire week. Time is recorded at a task level. Since the first report of this study in 1992, the number of tasks has been reduced through consolidation. Cataloging, for instance, is now divided into nine tasks rather than fourteen. Task consolidation makes data collection easier for staff and supports more meaningful analysis.

Tasks are organized into eight cost centers. Five are product centers, which create products and services: Acquisitions, Cataloging, Catalog Maintenance, Volume Preparation and Preservation, and a special project center, Conversion. The three other centers are overhead centers, which do not create products: Paid Leave, Automation, and Support Services. The latter two merit some explication.

The Automation Center includes the time of one staff member who provides information technology support for Technical Services. This includes management of servers, software and hardware ordering and installation, software application development (e.g., cataloger's workstation), and reengineering support. It also includes the time all staff spend learning to use general application software (mail systems, operating systems, word processing, etc.) and managing their personal computers. The Support Services Center includes all administration, meetings, professional activities, secretarial support, nondivisional work (such as materials selection or service on librarywide committees), and professional reading.

When participants self-report, there is always a potential for error. Yet there is really no way to control for error, because observation creates an artificial work environment that may not reflect normal work practices. Statisticians rarely recommend correcting for measurement error, because there is no way of knowing the error and any corrections may introduce other errors. Defining tasks clearly and making them reflect actual work processes makes record keeping for participants much easier and improves the chances of reliability. Also, data collection for this study has continued for more than ten years, and examination of the data shows results that reflect changes in library priorities. For instance, after a major serials cancellation program, the data showed increases in serials recataloging. Similarly when staffing was increased to support greater retrospective conversion, associated task time increased. The same is true for major system changes and upgrades; here the data show increases in training and documentation time. Finally, in the study we are not seeking precise data but rather more generalized data; thus staff are asked to estimate time spent at tasks, not to try to record it exactly.

Product vs. Overhead Centers

The division between product and overhead centers allows us to examine separately the time and cost of these different areas. Additionally, it allows layering on, by administrative levels, staff overhead costs to product center costs and demonstrates clearly the effect on product costs of staff time spent in paid leave, meetings, nondivisional work, professional and …