lib-s-mocs-kmc364-20141005043735 75 A Cost Effectiveness Model for Comparing Various Circulation Systems Thomas K. BURGESS: Washington State University Library Two models for circulation systems costing are presented. Both the auto~ mated and the manual models are based on experience gained in the analysis of circulation services at Washington State University library. Validation tests for the model assumptions are devised and explained. Use of the models for cost effectiveness comparison and for cost prediction are discussed and examples are given showing their application. INTRODUCTION Many methods for analyzing cost effectiveness have been presented recently in the literature.1 One main difficulty with studies of effectiveness is in quantifying the benefits, or in the case of libraries, assigning values to the quantity or quality of the services offered. 2• 3 One way to circumvent this difficulty is to compare the costs of different methods of providing the same services. Value assessment of the services is eliminated by keeping them constant as shown in most cost benefit studies.1- 6 This, of course, is not always possible when comparing manual library systems with mechanized systems. Library circulation systems, however, may fit this type of model with relative ease. For this reason, the models described below were developed to compare a manual with a mechanized system. They have the added advantage of allowing for the prediction of costs for either the manual or automated system based on certain circula- tion loads. The utilization of the models is probably best understood by working through an application. Therefore, a description of these applications as performed at Washington State University library will be used. Assump- tions based on practices peculiar to Washington State University are removed by the model through the use of the activities definitions for our library. Washington State University library has been operating a mechanized circulation system since 1967. Based on past experience, the system has recently undergone major modifications to improve its capabilities. We consider it to be a highly efficient machine circulation system. Thus, cost 76 Journal of Library Automation Vol. 6/2 June 1973 effectiveness comparison with a similar manual operation can provide infor- mation on effectiveness of automated circulation systems in general as well as on the WSU implementation. MODEL CONSIDERATION To insure that the comparisons were fair and that biases were held to a minimum, mathematical models had to be established with rather rigid constraints. Validations of these models had to be devised to insure that extrapolations of the model results were meaningful. Information about our manual system in operation prior to 1967 is sparse, as no analysis had been performed. It was decided that the manual model should, therefore, be a variant of the machine model, since our machine system includes a small manual system. If the models are to be useful to others, they should make very few assumptions about circulation tasks. Therefore, the models should break out each specific task so costs can be accumulated. This also insures that only circulation tasks are counted. If total hours of staff assigned to cir- culation are used as the basic labor costs, their time at other library func- tions are included and would provide erroneous data. Using a breakdown by tasks will allow use of the model even if major changes occur in orga- nizational or physical rearrangement of the circulation functions. Twenty-three basic activities were identified that would cover all cir- culation functions of our library. A similar list should be prepared for each library to be modeled.7 Our list can be used as a guide. These func- tions and their definitions are listed in Appendix A. Fifteen functions represent activity for which both the quantity of the activities and the average time to perlorm it are required information for building the model. Of the nine remaining activities, eight require only the measurement of total performance time. The last activity, computer operations, was sub- divided into three parts: computer charges, library equipment rental costs, and computer personnel costs. The computer personnel costs represent time donated by the Computer Center to keypunch, decollate and burst printouts, and prepare and schedule jobs. These personnel costs are a part of the machine system and are not reflected otherwise in the computer charges. These three charges are summed and used as a single dollar figure in the model. In our machine system as in many other circulation systems it is impossible to split our computer cost for each circulation sub- function because we use integrated data bases which are charged as a single storage rental cost and not split up among the various programs. The collection of data for this study could have resulted in a sizeable effort and could have unduly biased the data which were to be collected.8 • 9 For example, circulation clerks might have taken as much time to mea- sure the circulation transactions as the circulation transactions themselves required. Therefore, we requested supervisors to estimate the time neces- sary for these tasks, the number of transactions performed, and the per- A Cost Effectiveness Model/ BURGESS 77 centage of staff and student hours used. These data were developed monthly for a three month period during the middle portion of a semester. Validation of these estimates to insure their reasonability was accomplished by comparing the total time expended in circulation as reported in the col- lected data with the total time assigned to circulation activities as reported in the payroll records (the usual manner of estimating costs) .10 A sur- prisingly high degree of correlation was found primarily due to the fact that few of our circulation staff members have responsibilities outside of cir- culation. The payroll data also had to be adjusted to reflect actual hours used in circulation functions. A 25 percent figure was used: to reflect holidays and leaves-8.4-10 percent; coffeebreaks-8-12 percent; sickness-2 per- cent; tardiness and work slumps-3 percent; and miscellaneous-3 per- cent, for regular staff. By the same method 15 percent was determined for student help. The difference in total hours between the two samples · was less than 5 percent (Appendix B, Table 5). · The study data were collected from five separate organizational areas (three circulation desks, Technical Service Division, and the Library Ad- ministrative Office) which are reasonably independent of each other; the monthly variation in activities reported by the various units was also closely correlated. For example, the percent increase in checkouts for a month was approximately the same at all three circulation desks. MODEL CALCULATIONS-AUTOMATED SYSTEM Mtmthly totals were averaged for each activity's transaction time, number of transactions, and percentage of effort allocated to staff, or student labor . (Appendix B, Tables 1, 2, and 3). Average hourly wages were developed separately for student (part-time help) and staff based on salaries of personnel allocated to circulation. The total hours and salarjes were then calculated for staff and student help for each activity. The follo~g ex- ample shows the formulas used in calculating some of the entries in; tpe tables: A1 Manual Checkout Transactions ( AlT) times transaction time ( AlTT) equals total time expended ( AlTE) adjusted to hours [I] Total part-time help in hours (AhTH) equals (AITE) times the per- centage of student effort ( AIPPTH) [2] Total staff hours ( Al 8 ) equals ( AlTE) times the percentage of staff help (AlPs) . [3] Total salaries (AITs) equals (AlPTn) times student rate (RPTH) plus . (Al8 ) times staff rate (Rs) [4] X Shelving Total student hours XPTn = XTE · XPPTH [5] 18 Journal of Library Automation Vol. 6/ 2 June 1973 Total staff hours Xs = XrE · XPs Total salaries XTs = Xs · Rs + XPTH · RPTa All other activities were calculated in the same manner as shown above. Personnel hours used were totaled and multiplied by the hourly rates. The salary totals and the computer costs were then added together to get the total system cost per month. (Appendix B, Table 6) v v [6] [7] Total salary cost = 1.15 ~ iPTa · RPTa + 1.25 1: is · Rs [8] i=A i=A Figure 1 represents curves of monthly cost vs. monthly circulation. The automated system curve was determined from the initial model plotted point, and from extrapolations to other plotted points which were computed b~sed on the following factors: A 25 percent increase or decrease in cir- culation will result in a 5 percent increase or decrease in computer costs. This estimate results from analyzing the computer processes. The bulk of the computer cost results from sorting and other total file processes which are reasonably insensitive to changes in volume of updating. A factor of 25 percent change in circulation results in a 30 percent change in personnel costs. The 5 percent differential may be conservative, but results from the need for additional supervisory support with its higher salary for each additional operational position added. Using the above factors, several additional points were predicted and plotted and the automated system curve was drawn to fit these points (Appendix B, Table 7). Validation of these factors was determined by using budget information and circulation data available from the year 1968 (Appendix B, Table 8). These data were used to establish a point on the graph. The 1968 costs were compared to the predicted cost as shown by the curve for the circulation volume in 1968. This provided a cost differential which was within 1 percent of the curve predicted costs (Fig- ure 1) . . These data were adjusted to reflect annual circulation hours used in circu- lation in 1971. MODEL Cl\.LCULATIONS-MANUAL SYSTEM · The manual model was a modification of the automated model. Obviously no machine costs were incurred, but costs for filing and retrieving cards from large tub files of book cards of items in circulation must be added to each' check-out or check-in procedure as well as to snags, holds, and other categories. Since ·some· loaned materials are not included in our automated system, a small manual circulation operation runs parallel to the automated system and was included in the automated study. This small manual system served as the base activity for the manual model in the study. A Cost Effectiveness Model/BURGESS 79 Retrieval time from a card tub file is dependent on the number of the cards in the file. The tub file size is approximately equal to the size of the computer's circulation file. Sample filing times were made on a catalog card file of comparable size. The results were an average of 40 seconds per item on timings of single records and of batches of alphabetized rec- ords to be filed. This figure was then used to extend the average time of the appropriate activities in the parallel manual systems data (Activities Al, B, Il, K, M, and Q). Following the calculation method used in our automated system, data were developed from the 1971 circulation data and a curve was drawn for the manual system (Appendix C and Appendix B, Table 7 and Figure 1). Validation of this curve by budget information available from 1967 shows that the difference in the predicted cost from actual cost was less than 2 percent (Appendix C, Table 5). This represents a significant corre- lation and validates the entire manual model. GENERALIZED USE OF THE MODELS As has been shown by the example of its use at WSU, the models pro- vide for two functions: cost comparison of automated and manual circula- tion systems at the same levels of book circulation, and prediction of cost in either a manual or automated system at different levels of circulation. Of course, combinations of these models may be made, such as: at what circulation levels are costs of both models equal? Or, what will my costs .e • Cron-over point 4,000 8/YJO 12,000 16,000 20/)00 24,000 28,000 32,000 36,000 Monthly