Time-Driven Activity-Based Costing for

Inter-Library Services: A Case Study

in a University
by Eli Pernot, Filip Roodhooft, and Alexandra Van den Abbeele
Available online 12 July 2007

Although the true costs of inter-library loans INTRODUCTION

(ILL) are unknown, universities increasingly rely Ever diminishing university funding by the government1–4
on them to provide better library services at urges university library management to provide better library
services at a continuously lower cost.2 To cope with these cost
lower costs. Through a case study, we show how pressures, library management needs to improve its under-
to perform a time-driven activity-based costing standing of the relevant cost drivers.
analysis of ILL and provide evidence of the To date, most studies on university library costs have relied on
traditional activity-based costing (ABC) systems.2,3,5 Although
benefits of such an analysis.
these studies found ABC both providing library management with
a more detailed cost analysis of activities and services, and
assisting management in understanding what actually causes
certain costs,2,3 ABC also appeared to cause two significant
problems. First, setting up an ABC system can be very costly,
especially if the current accounting system does not support the
collection of ABC information. Second, the system needs to be
regularly updated, which further increases its cost.3 These
limitations motivated Kaplan and Anderson6 to develop time-
driven activity-based costing (TDABC), a revised version of
ABC, solving these problems, without losing the benefits. The
most important characteristic of this technique is its simplicity, as
only two kinds of parameters need to be estimated: the number of
time units (e.g. minutes) consumed by the activities related to the
cost objects and the cost per time unit. Hence, TDABC systems
cannot only be implemented quicker (and thus cheaper), but also
updated easier, than traditional ABC systems.6
Since services constitute the ideal setting to implement a
TDABC system in order to improve the service’s cost manage-
ment, we performed a case study at the Arenberg library of the
Catholic University of Leuven (KULeuven). Although we argue
that TDABC could improve the cost management of all library
Eli Pernot is aspirant of the Fund for Scientific Research, services, as identified by Ellis-Newman and Robinson,2 Ellis-
Flanders. Katholieke Universiteit Leuven, Newman,3 and Goddard and Ooi,5 we only concentrated on the
Naamsestraat 69, 3000 Leuven, Belgium inter-library loan (ILL) service, for several reasons. First, ILL is
[email protected]; directly linked to the increasing cost pressures on university
Filip Roodhooft, Katholieke Universiteit Leuven, libraries. Indeed, Nitecki and Renfro7 confirm that library
Naamsestraat 69, 3000 Leuven, Belgium management relies more and more on ILL, in order to provide
Vlerick Leuven Gent Management School, good library services in times of decreasing library buying
Leuven Campus Vlamingenstraat 83 power. When books or journals are not available in the library of
3000 Leuven, Belgium choice, readers can request them from other libraries via the two-
[email protected]; directional ILL system. Second, an accurate cost calculation is
Alexandra Van den Abbeele, Katholieke Universiteit Leuven, important for ILL services. On the one hand, the processing costs
Naamsestraat 69, 3000 Leuven, Belgium for an inter-library loan are higher than for an internal loan,
[email protected]. because inter-library loans require higher-level staff and are

The Journal of Academic Librarianship, Volume 33, Number 5, pages 551–560 September 2007 551
more time consuming to process.3,8 On the other hand, also the Technology (NIST) Library and revealed a significant increase
costs of different ILL requests vary significantly, with a huge in the total average cost per request, from $29.91 in 2003 to
difference between straightforward requests and the most $41.20 in 2004.19
difficult ones. Hence, allocating the same amount of costs to We would like to contribute to this ILL cost study literature by
these different types of requests would be inaccurate.3 Third, providing a whole new approach, based on time-driven activity-
contrary to overall library costs, specific ILL costs have been based costing (TDABC), for calculating ILL costs. Contrary to
extensively studied in the past, frequently as one of the four traditional ILL cost studies such as the ARL studies, TDABC
general agreed ILL evaluation criteria: fill rate, turnaround time, does not follow Lor’s20 ‘‘macro-approach’’ of dividing yearly
user satisfaction and cost.9 In the following paragraphs, we aggregated costs by total number of requests filled, but the
summarize the findings of some relevant studies. For a more ‘‘micro approach’’, involving cost data collection and calcu-
extensive overview of ILL cost studies, we refer to Stein.9 lation for the various steps in ILL processing. This way, the
In 1992 the Association of Research Libraries (ARL), TDABC approach results in disaggregated costs for all ILL
conducted the first major ILL cost study in North America service activities, instead of aggregated dollar figures like $x for
and found that research libraries spent on average *$18.62 to borrowing and $x for lending. As the KULeuven library
borrow (ranging from $9.84 to $30.27) and $10.93 (ranging manager confirmed that ‘‘ILL is extremely time consuming,
from $6.92 to $17.49) to lend via ILL.10 The follow-up study11 difficult to define in terms of subactivities, significantly different
by ARL expanded the research to all four ILL performance across the requests and costly’’, the in-depth TDABC analysis
measures and increased the library sample from seventy-six to really answers a practitioner’s request in order to reveal the true
ninety-seven research libraries. ILL costs appeared to have cost of ILL and clearly show what activities cause the costs.
declined slightly to $18.35 for borrowing and $9.48 for lending. The remainder of this paper is organized as follows. First,
Additionally, twenty-two college libraries were questioned, that we present the inter-library TDABC case. Second, we discuss
showed significantly lower ILL borrowing costs of $12.08 and the management implications resulting from our analysis.
lending costs of $7.25.11 The ARL methodology was sub- Third, we end the paper with a brief conclusion.
sequently adapted for use in Australia and the Nordic countries
(Denmark, Finland, Iceland, Norway and Sweden). In Australia, TIME-DRIVEN ACTIVITY-BASED COSTING FOR
the average borrowing unit cost was AU$32.09,12 while INTER-LIBRARY LOANS
research in the Nordic revealed an average borrowing unit cost In this part of the paper, we thoroughly explain the inter-library
of o12.56 and average lending cost of o7.24.13 TDABC analysis. To that end, we first briefly describe ILL at
The second ARL performance measurement study14,15 of the KULeuven Arenberg library. Then, we present a thorough
seventy-two North American research, college, and govern- activity analysis and derive time equations for each request
mental libraries also measured patron-initiated (or unmediated) activity. Third, we provide an overview of all related ILL costs
ILL criteria and found lower unit costs than for mediated ILL and identify the different costs per minute. Based on these time
services. Borrowing unit costs for user-initiated ILL ranged equations and costs per minute, we finally calculate the cost of
from $2.39 to $14.7, compared to $17.50 for mediated ILL. different outgoing and incoming requests.
Similar results were found for lending, with a lending unit cost The activity data for this study were gathered through direct
for user-initiated ILL ranging from $3.27 to $12.06 and $9.27 observation and multiple interviews with both the library
for mediated ILL.14,15 Yet, the most famous consortium manager and the ILL responsible (cf. Goddard and Ooi5). Cost
regarding patron-initiated ILL is undoubtedly OhioLINK, a data were obtained from the library accountant and other
consortium of Ohio-based college and university libraries, information was found by means of archival research, which
which created an alternative for traditional ILL. By utilizing included inter alia the annual report. All observations were
user-initiated requests, a single automated system for fifty-two made during 2006 and all financial data (like wages and
library collections, a commercial delivery service, and a overhead costs) are from 2005.
staffing context that requires little more than low-cost student
Inter-Library Loans (ILL) at the KULeuven Arenberg
labor, OhioLINK has achieved average ILL unit costs of not
Library
more than one dollar. Therefore, OhioLINK members made
ILL affordable on a mass volume basis and do not even call it The KULeuven Arenberg library’s mission is ‘‘to develop a
ILL any longer, but simple circulation.16,17 well-balanced collection for education and research, with
A similar patron-initiated borrowing service project was extended opening hours for students, an electronic library that
implemented by seven private East Coast institutions and can be consulted at individual workplaces and the facilities for
called ‘‘Borrow Direct’’. Compared to traditional ILL costs, guided self-education’’ (Annual Report, 2005). The recent
Borrow Direct achieves significantly lower per-transaction trends to automate standard processes, like lending and
costs, namely $5.33 for borrowing and $6.48 for lending.7 receiving books, and to digitalize the library are aimed at
Nevertheless, these interlending costs remain higher than attaining this goal at the lowest possible cost. Furthermore, the
OhioLINK costs, primarily because Borrow Direct members library is a member of the LIBIS-network, which manages the
do not share a common automated system and must operate library systems of the entire KULeuven and LIBIS-Net. LIBIS-
across state lines.17 The same cost comparison can be made for Net’s mission is to support participating libraries in their library
the Jefferson County Public Library (JCPL), member of the management and the services they provide, such as ILL. Like
Prospector INN-Reach consortium in Colorado, although the any other ILL system, LIBIS-Net. supports on the one hand
ILL per-transaction costs of Prospector appeared much closer that the KULeuven library borrows books and requests
to the OhioLINK costs: $1.17 for lending and $1.56 for journal articles that are not present in its collection, the so-
borrowing.18 Another very recent ILL cost investigation was called outgoing requests, and on the other hand that the library
conducted at the US National Institute of Standards and lends items to other libraries, the so-called incoming requests.

552 The Journal of Academic Librarianship

Although ILL basically only refers to ‘‘loans’’, we will use the Figure 2
term in the remainder of this paper to refer to both books loans Flows of Books and Journal Articles
and article deliveries. for Incoming Requests
Fig. 1 depicts the relations between the three parties in-
volved in an outgoing request, which generally consists out of
three flows.
Only when a book is not returned within the allowed
borrowing period, the library sends a reminder for return,
creating a fourth flow (4). Obviously, when the request concerns
a journal article, the ILL process stops after flows 1 and 2,
because a PDF file or a hard copy must not be returned. For most
Belgian libraries, payment for ILL is handled by the clearing-
house system of Impala, comparing earnings and dues on a 1 to 1
basis. While libraries with a negative balance receive an invoice
from Impala, those libraries with a positive balance may send the
clearing-house an invoice. When payment is not handled by
Impala, the KULeuven library either receives an invoice from
the other library or settles the payment by means of an IFLA
(International Federation of Library Associations) voucher, the
latter being utilized more and more frequently for international
ILL services. Obviously, the KULeuven library charges the provided an estimation of the time needed to perform this
reader for the ILL costs. All financial transactions follow flows subactivity once. Based on these estimations, we derived eight
(2) and (3) in Fig. 1, but do not have to coincide with the physical time equations, one for every main activity of an outgoing and
flow of the book or article. incoming request, which can be found in Table 1.
From the perspective of the KULeuven library, incoming The interpretation of these time equations is similar for all
requests involve only two parties and consist of three flows equations. The total time needed – in minutes – for a certain
when the ILL concerns a book loan, represented in Fig. 2. activity is the sum of the times needed to perform a certain
Activity Analysis and Time Equations subactivity, taken into account that certain request character-
istics, like the fact whether the request concerns a book or
To construct appropriate time equations, we performed a journal article, influence the presence of certain terms in the
thorough activity analysis of ILL at the Arenberg library. We equation. For that reason, these characteristics are included by
identified four common ILL activities (process, deliver, close dummy variables, which are equal zero or one, the latter when
and invoice), each implying several subactivities, which differ the request actually shows this characteristic and its time needs
for outgoing and incoming requests. For each of these to be included in the time calculation. The role of these dummy
subactivities, the ILL responsible at the KULeuven library variables needs to be emphasized, as these variables are crucial
in understanding the utilization of the time equations.
Furthermore, the time equations are set up chronologically,
Figure 1 implying that both the main activities and the subactivities
Flows of Books and Journal Articles occur in the order of the position in their time equation. We will
for Outgoing Requests now discuss the different equations and indicate which
subactivities correspond to the times in the equations.
Outgoing Request
An outgoing request starts with processing the request. This
activity always requires library staff to receive the request either
personally or electronically, to select a library possessing the
requested book or article, to send the request to this library, to
make a print-out, to enter data in an Excel-file, and finally to
classify all print-outs (6.8 min). Besides these activities, the
request potentially includes two extra activities, increasing the
time needed to process the request. First, the reader may ask
library staff feedback via e-mail, telephone, or at the library desk
(6.3 min). Second, the other supplying library may negatively
respond, implying that the Arenberg library staff has to start over
the process (6.6 min). Sometimes, in case of a rare document that
is difficult to find, this process may need to be repeated several
times. In that case, this term in the time equation has to be added
more than once as well. Furthermore, we want to remark that the
actual time equation does not take into account the experience of
the different library employees working on ILL.
The second main activity of an outgoing request concerns
the delivery and is substantially different for an article in PDF,

September 2007 553

 Table 1
Time Equations Per Main ILL Activity
Type of ILL Request Main Activity Total Time Needed Per Activity (In Minutes)a
Outgoing request
Process 6.8 + (6.34feedback) + (6.64negative)
Deliver (3.24article-in-PDF) + [(3 + 3b + (2.44successful) + (14unsuccessful))4article-on-paper] +
[(6.3 + (54prolongation))4book]
Close (0.74article) + [(3.5 + (24reminder))4book]
Invoice [(0.8 + 0.2c)4supplier-invoice] + (0.54IFLA) + (3.54no-client-invoice) +
[(0.8 + 0.2d)4client-invoice]

Incoming request
Process 3.2 + (6.34feedback) + (24negative) + (4.44article-on-paper) + (2.54book)
Deliver (24article-in-PDF) + [(3.4 + (2.44successful) + ((4 + (2.44second-successful) +
(14second-unsuccessful))4unsuccessful))4article-on-paper] + [(5.1 + (44reservation))4book]
Close (0.74article) + (2.24book)
Invoice [(0.8 + 0.2e)4client-invoice] + (0.54IFLA)
a
The total time needed per activity is the sum of the time equation terms, that are either a number of minutes without multiplication (e.g. 6.8 min for process of outgoing request)
or a number of minutes multiplied by a dummy variable (e.g. 6.64negative for process of outgoing request). These dummy variables are set to either zero or one, the latter when
this specific request characteristic is present.
b
When the supplying library delivers the requested article on paper, this article is first received (3 min) by library staff and then scanned (3 min). Both subactivities and their times
needed are separated in the time equation, because they are subject to a different cost per minute, as made clear in Table 2.
c
When the KULeuven library pays the outgoing request by a supplier-invoice, library staff has to check the invoice and adjust the electronic logs, taking up 0.8 min per request.
Then the accountant actually books the invoice (0.2 min). As the accountant is more expensive than other library staff, both subactivities are separated (cf. Table 2).
d
When an outgoing request is paid by a client-invoice, library staff has to check the invoice and adjust the electronic logs, taking up 0.8 min per request. Then the accountant
actually books the invoice (0.2 min). As the accountant is more expensive than other library staff, both subactivities are separated (cf. Table 2).
e
When an incoming request is paid by a client-invoice, library staff has to check the invoice and adjust the electronic logs, taking up 0.8 minutes per request. Then the accountant
actually books the invoice (0.2 minutes). As the accountant is more expensive than other library staff, both subactivities are separated (cf. Table 2).

an article on paper or a book, as shown with dummy variables perform the payment. The reader pays the ILL fee at the library
in the time equation for this activity. Journal articles are sent desk in advance (3.5 min). Yet, if the reader is connected to a
electronically when possible. Library staff then receives the university department, a client-invoice is made based on the
article via e-mail, temporarily saves the article in the PDF Excel-file data (0.8 min). These data are handed over to the
database, and sends the paper to the reader via e-mail (3.2 min). accountant, who draws up the invoice and follows up on the
However, when the article is received on paper via postal mail payment (0.2 min).
or shuttle service (3 min), it is scanned into a PDF file (3 min)
and, if successful, temporarily saved and send via e-mail (2.4 Incoming Request
min). When the scan was unsuccessful, the original hardcopy Similar to outgoing requests, incoming requests start with
of the article on paper is put in an envelope and sent via postal processing, which involves printing the request, entering data
mail (1 min). Obviously, a book is always received physically in an Excel-file, classifying the print-outs, and searching books
through postal mail or shuttle service, after which the user is and journal articles in the LIBIS-catalogue and e-sources
notified via e-mail and a copy of the request is stored (6.3 min). subscriptions (3.2 min). If the Arenberg library does not
However, if the reader requests a prolongation on the book, possess the requested item, library staff negatively responds to
library staff has to contact the supplying library, ask a loan the requesting library (2 min). Then, library staff searches the
period extension, and mail the answer to the reader, requiring shelves, which takes up 2.5 min for a book and 4.4 min for a
an extra 5 min. journal article, because journal volumes need to be checked.
After delivery, library staff closes the request in LIBIS and Obviously, the reader or requesting library may also ask
the Excel file, which takes 0.7 min for an article and 3.5 min feedback via e-mail, telephone or at the library desk (6.3 min).
for a book, because the latter also needs to be sent back to the Secondly, the requested article or book has to be delivered.
owner library by postal mail or shuttle. When the book is not A PDF file of a requested journal article is sent by e-mail to the
returned in time, a reminder is sent to the reader (2 min). requesting library (2 min). An article on paper is scanned, by
Finally, invoicing is usually dealt with by the clearing house one or more students working part-time at the library (3.4 min).
of the Impala system. In most other cases, the KULeuven If successful, library staff saves the PDF file and sends it via e-
library receives an invoice, which is checked, entered in the mail (2.4 min). Otherwise, the article is copied, subsequently
Excel file (0.8 min), and subsequently dealt with by the scanned (4 min), and delivered, meaning either a second
accountant (0.2 min). If the supplying foreign library is paid successful PDF version via e-mail (2.4 min) or the hard copy
with an IFLA voucher, library staff only needs 0.5 min to via postal mail or shuttle service (1 min). A book is packed,

554 The Journal of Academic Librarianship

made ready for transportation, and sent, after which its status is copying and scanning activities for which a specific cost of
changed in the log file (5.1 min). If a reservation is made for a o0.02 per minute is derived, covering the scanner/copier and all
book that has been borrowed, library staff will process the related costs, such as paper and maintenance. Finally, we also
reservation with a message sent in LIBIS (4 min). distinguish a specific shipping cost of o0.02 per minute.
Concerning closure, library staff will close an article request Based on these separate cost rates, we identify five activity
in LIBIS and the Excel file (0.7 min). After a lend book is cost rates. The first cost rate is the fully loaded staff labor cost
received back, this request will also be closed in both LIBIS and of o0.36 per minute, including labor costs and general
the Excel file (2.2 min). overhead costs. In case the ILL request involves sending a
Because invoicing between the Arenberg library and other book or article on hard copy, we add a specific shipping cost of
Belgian libraries is almost always done via the Impala system, o0.02 per minute. Third, the fully loaded student labor cost
only a minority of incoming requests lead to an invoice, rate equals o0.31 per minute, covering labor cost (o0.28) and
handled by the same workflow as the client-invoice of an overhead (o0.03). Next, the accountant costs o0.46 per
outgoing request made for the requesting department (0.8 min minute, based on his labor cost (o0.43) and the overhead cost
from library staff and 0.2 min from the accountant). Yet, if a (o0.03). Finally, the machine cost rate for scanning and
requesting foreign library pays with an IFLA voucher, library copying equals o0.02 per minute. These five cost rates will be
staff only needs 0.5 min to handle the payment. used to calculate the different costs for different requests in the
following paragraph and are summarized in Table 2.
Identification of the Cost Per Minute
In this part of the TDABC analysis, we identify how much Cost Calculation of Inter-Library Loans
one time unit – 1 min – of every activity, defined in the In this part, we derive two cost tables to calculate the cost of a
previous step, actually costs. All cost rates were calculated by specific ILL request, one for outgoing and one for incoming
means of data provided by the library manager. According to requests. Yet in order to facilitate the cost calculation, we
him, the total number of personnel assigned to ILL represents explicate a second distinction, namely the one between journal
2.5 full time equivalents. With a theoretical capacity of 38 h a articles and books. Obviously, this distinction was already
week, a total of 95 h would be available for ILL activities per implicitly present in the time equations, by means of dummy
week. Yet, instead of using this utopian kind of capacity, we variables. However, as our cost tables do not contain dummy
base our calculations on the more realistic practical capacity, variables, this distinction is needed to add all appropriate costs.
which is set at 80 percent of the theoretical capacity, a standard Furthermore, we also distinguish between articles on paper and
number in the management accounting literature (e.g. Kaplan articles in PDF, in order to increase the clarity of our cost tables.
and Anderson6) and also the number given by the ILL Table 3 shows the cost table for the three types of outgoing
responsible. Therefore, the practical capacity for ILL activities requests.
is 76 h or 4560 min per week. This table can be used to calculate both the total time needed
to handle the request and the cost of all activities related to the
request. To this end, one first identifies whether the outgoing
request deals with a book, article available in PDF, or article
‘‘In... [time-driven activity based costing] available on paper. When the appropriate main column is
analysis, we identify how much one time unit - selected, the rows of this type of outgoing request must be read
one minute - of every activity, defined in the from top to bottom. Every row in the table represents one
activity and contains the activity description, time needed (in
previous step, actually costs.’’ minutes), and cost incurred (in o). The activity description in
the first column and time in the second column result from the
different time equations in step one of the analysis. The third
The 2.5 FTE consist of six people, each working 100 percent column indicates the cost of performing an activity once and
(the ILL responsible), 75 percent, 37.5 percent, 25 percent, 6.25 results from multiplying the time needed by the cost per minute
percent, and 6.25 percent of their time on ILL. Based on their for that activity, identified in step two of the analysis (cf. Table
different gross wages, the total labor cost of these people is 2). Obviously, we are mainly interested in these costs in order to
o1,521.56 per week. Dividing this amount by the practical calculate the total cost. Yet, the column with the number of
capacity of 4560 min per week, results in a cost of o0.33 per minutes per activity is included in the cost table to allow
minute. Besides these employees, the Arenberg library hires pellucid calculation of the total handling time needed. Ulti-
students to perform scanning and copying activities for
incoming requests. Based on their hourly wage, we calculated
their labor cost per minute as o0.28. Finally, the Arenberg Table 2
library also employs an accountant, who books invoices to Costs Per Minute
readers and deals with the potential payments of supplying and
requesting libraries. His monthly gross wage is o3,117.11, Cost Type Cost Per Minute
implying that his cost per minute amounts to o0.43. Staff labor cost 0.36
According to the Arenberg library accountant, the ILL Staff labor cost + shipping cost 0.38
overhead costs concern equipment and daily working costs. In
the same way, so by dividing total costs by total number of Student labor cost 0.31
minutes (4560 per week), we calculated a general overhead cost Accountant labor cost 0.46
of o0.03 per minute, covering furniture, computers, computer
Scanning machine cost 0.02
supplies, and paper. In addition, article requests can require

September 2007 555

 Table 3
Outgoing Request Cost Table

I
To calculate the total cost of an outgoing request, one first identifies whether the request deals with a book, article available in PDF or article available on paper. That way, the appropriate
column is selected, which must be read from top to bottom. Every activity is represented by one row in the table, specifying activity description, time (in minutes) needed and costs (in o)
incurred. The total cost of an outgoing request is the sum of the two subtotals. The calculation of the total time needed is similar to the cost calculation.
II
The subtotal of the cost of ‘‘standard activities’’ is the sum of costs of the included activities. For every request type, one standard activity involves a choice. Depending on whether the
request requires a client-invoice, the appropriate cost, separated by ‘‘or’’, needs to be included in the subtotal. The separation for the client-invoice alternative is a consequence of a different
cost per minute (0.36oand 0.43o) for the two related subactivities (cf. Tables 1 and 2).
III
For an article on paper, one other standard activity involves a choice. Depending on whether the scan of the article on paper was successful, the PDF file is saved and sent or the hardcopy is
sent, indicated by ‘‘or’’ in the table. In order to calculate the total cost correctly, this request characteristic needs to be identified, so that the appropriate cost is chosen from this table row.
IV
The subtotal of optional activities is also the sum of the included activities, taking into account that all costs are added only when the characteristic is present.

mately, the total cost of an outgoing request is the sum of two taking into account that all costs are added only when the
subtotals, visualized by two sections at the left hand side of the characteristic is identified to be present.
table. So to calculate the total cost of a certain outgoing request,
The first subtotal of ‘‘standard activities’’ is the sum of the we first need to identify the type of request – book, article in
costs of the activities in this section of the table. For each of the PDF, or article on paper – and the characteristics determining
three request types, one of these standard activities involves a the occurrence of certain activities. Then, we sum the costs of
characteristic identification. Depending on whether the request the identified activities into two subtotals. Finally, these
requires a client-invoice or not, the appropriate cost, separated subtotals are added into the total cost. Let us explain this
by ‘‘or’’ in the table, needs to be included in the subtotal. For approach by a thorough example.
an article on paper, another identification has to be made. Assume that the KULeuven library receives the request
Depending on whether the scan of the article on paper was from a reader to order a certain journal article via ILL. The
successful, the PDF file is saved and sent or the hardcopy is reader did not find the journal article in the library collection
sent, indicated by ‘‘or’’ in the table. In order to calculate the and asks for help at the library desk. After he hears about the
total cost correctly, these request characteristics need to be possibility of ILL, he fills in a request form and says he
identified in beforehand, so that the appropriate costs are prefers an invoice to payment in cash. Yet, the journal article
chosen from the table. The subtotal of ‘‘optional activities’’ is appears difficult to find, as the first supplying library
also the sum of the included activities in this table section, negatively responds to the ILL request. In addition, the

556 The Journal of Academic Librarianship

second supplying library sends the article on paper, so that a fourth row. After closing (o0.25), this request is paid via an
scan is performed. Yet, this scan appears unsuccessful, so that invoice (o0.29 + o0.09). Therefore, the subtotal of standard
the hardcopy of the article is sent by postal mail. Finally, the activities amounts to o5.68. Additionally, this request is
foreign supplying library requests payment by means of an hampered by a first negative request (o2.38) and requires
invoice. feedback (o2.27), as well as a supplier-invoice, since the
In order to calculate the total cost of this request, we first payment is not handled by Impala (o0.29 + o0.09). This
need to identify its type and characteristics. It is clear that this leads to a second subtotal of o5.03 for optional activities.
request concerns an outgoing request of an article that is Hence, the total cost to handle this outgoing request amounts
delivered on paper by the supplying library. That means that to o10.71. In the same way, the subtotal of time needed for
we consider the third main column of Table 3 and need to standard activities adds up to 15.5 min. As the optional
include the standard activities of processing (o2.45), receiv- activities require 13.9 min, the total time needed to handle
ing (o1.08), and scanning (o1.14) in the addition of the first this outgoing request is 29.4 min.
subtotal. Furthermore, we know that the article is scanned The cost table for incoming requests, shown as Table 4, is
unsuccessfully, so that the hardcopy is sent to the reader by very similar to the one for outgoing requests and can be used
postal mail (o0.38). Hence, we select this activity of the accordingly.

Table 4
Incoming Request Cost Table

X
To calculate the total cost of an incoming request, one first identifies whether the request deals with a book, article available in PDF, or article available on paper. That way, the appropriate
column is selected, which must be read from top to bottom. Every activity is represented by one row in the table, specifying activity description, time (in minutes) needed and costs (in o)
incurred. The total cost of an incoming request is the sum of the three subtotals. The calculation of the total time needed is similar to the cost calculation.
XI
The subtotal of ‘‘process’’ includes the cost of the standard activity process and the optional cost of a negative response. If the KULeuven library actually sends this kind of negative
response, no other activities than process occur, implying that this subtotal becomes the total cost.
XII
The subtotal of ‘‘other standard activities’’ is the sum of the costs of the included activities.
XIII
For an article on paper, one other standard activity involves a choice. Depending on whether the scan of the article on paper was successful, the PDF file is saved and sent or the article is
copied and a second scan is made, indicated by ‘‘or’’ in the table. In the latter case, a second choice is needed: if the second scan was successful, the resulting PDF file is saved and sent;
otherwise the hardcopy is sent. In order to calculate the total cost correctly, these request characteristics need to be identified, so that the appropriate cost is chosen from this table row.
XIV
The subtotal of ‘‘optional activities’’ is also the sum of the included activities, taking into account that all costs are added only when the characteristic is present.

September 2007 557

 Two differences with the outgoing request cost table worth investigate the possibility of buying a more professional
mentioning are the following. First, this cost table includes an scanner. This kind of scanner would not only lower the
extra subtotal, named process, which separates the standard scanning time, but also augment the quality of the first scan,
activity of process from the other standard activities (the thereby reducing the possibility to need a second scan. One
second subtotal). The reason for this approach is the possibility could, of course, also consider abandoning the approach of
that the KULeuven library sends a negative response to the scanning articles on paper, as this activity consumes relatively
requesting library. In that case, no other standard activities than much time and appears unsuccessful in more than half of the
process occur for this request, implying that this subtotal requests. Yet, this recommendation was considered incompat-
becomes 5.2 min or o1.87. Obviously, this cost, potentially ible with the fundamental mission of the Arenberg library,
supplemented by a cost of o2.27 for providing feedback, is a which is to offer the best possible service. Concerning ILL, this
complete loss for the KULeuven library, as it does not receive a service implies that the library wants to deliver the requested
fee for delivering a negative response. Yet, if the library does articles as soon as possible. As an article in PDF can be sent by
find the requested book or journal article in its collection, e-mail instead of postal mail, the mission enforces scanning
negative is not selected. Therefore, this subtotal amounts to articles. Furthermore, the library manager indicated that readers
o1.15, which has to be added to the resulting subtotal of other benefit from getting an electronic version instead of a hard
standard and optional activities. Second, an article on paper copy, as the first is more easily stored and retrieved. Finally, the
involves a second choice in the other standard activities PDF files also offer the possibility to be re-used by the
subtotal. As in the case of an outgoing request, the requested Arenberg library. After the article is successfully scanned, the
article found on paper is scanned, so that it can be sent PDF file is stored in the PDF database, so that it can be copy-
electronically. Hence, the first choice in the cost calculation is pasted for future requests of the same article.
the same, namely whether the scan is successful. Yet, if the Second, we notice that the search activity requires approx-
scan appears unsuccessful, the article is not immediately sent imately 75 percent of the process time, depending on the
on hardcopy, but copied so that a second scan can be made of specific characteristics of the request. In addition, we wonder
the copied version. If this second scan is successful, the whether the Arenberg library staff actually adds any value with
resulting PDF file is saved and sent; otherwise the hardcopy is this activity, meaning: is the reader not able to perform this
sent. Hence, in the incoming case a second choice is needed, search his- or herself? If so, the library manager can consider
namely whether the second scan is successful, which implies outsourcing this activity to the reader, by improving and
the addition of more time and cost and is indicated by ‘‘or’’ in simplifying the search process via the Internet. Hence, the
the table. search costs identified in our analysis form the base against
which the related process change costs need to be compared.
THE BENEFITS OF TDABC: SOME MANAGEMENT Third, the library should consider improving the activity of
IMPLICATIONS providing feedback, as this potential request characteristic takes
In order to adequately assess the ILL costs of the Arenberg up o2.27 of the o8 ILL fee. A possible process improvement
library, we need to compare those costs with the ILL would be to set up an electronic portal with a ‘‘Frequently
‘‘earnings’’, the price which the library charges to the reader Asked Questions’’ section and the possibility to post other
or requesting library. This price is fixed at o8. Although this questions. This way, readers could find the answers to
price covers the costs of simple requests, this amount will not previously asked questions, which are already in the database,
cover more complicated requests with several options, like the themselves. At the same time, library staff could easily add
outgoing article on paper request (o10.71), described in the answers to new questions, still being raised via the portal. The
previous paragraph. Outgoing request characteristics like cost of this portal needs to be weighted against the cost of
feedback, potential negative response from the supplying providing feedback individually.
library, payment in cash, reminder, and prolongation in case
of a book and scanning in case of an article on paper, all
significantly increase the cost of this type of request. Some
characteristics are less prevalent for incoming requests, ‘‘TDABC [Time-driven activity based costing
although feedback, reservation, and especially scanning remain
resource consuming characteristics.
analysis] offers the possibility to visualize which
Therefore, we can conclude that ILL earnings often do not activities consume the largest amount of time
cover ILL costs. The TDABC approach, with its time equations, and are thus most suitable for amelioration, in
offers some specific advantages, as these equations create larger order to reduce the total ILL service cost.’’
transparency than a traditional cost accounting system or an
ordinary ABC system. The time equations clearly show which
activities demand more time, and thus lead to higher costs. This
way, the library manager can take appropriate actions to lower The main point of this discussion is the fact that these
the time needed to handle a certain request. suggestions for improvement show the strength of the analysis,
First, the time equations show that scanning and copying an as they are all based on the time equations and subsequent cost
article on paper consume a relatively large amount of time. calculations. TDABC offers the possibility to visualize which
This activity requires 3 min for an outgoing request, while an activities consume the largest amount of time and are thus most
incoming request demands a scanning time between 3.4 and suitable for amelioration, in order to reduce the total ILL
7.4 min. Discussions with the library manager revealed that service cost.
scanning articles is a time consuming activity, because the Although the previous paragraph exclusively deals with
scanner is outdated. Hence, we argue that it is worthwhile to costs, there exists a second way to improve the balance

558 The Journal of Academic Librarianship

between ILL costs and earnings, namely altering the price for shows how the time-driven cost accounting technique works in
ILL. Based on the time equations, it becomes clear that when practice, by means of clear time equations, cost tables, and
library staff tries to speed up the delivery, it increases the cost several recommendations, but also exemplifies the additional
as well; think for example of the scanning activity in the case of benefit of disaggregated time-driven activity-based costs
an article on paper. Yet, one can imagine that some readers compared to traditional ILL cost studies. Therefore, we argue
value this service more than others, and thus are prepared to that this cost accounting technique is very suited to cope with
pay extra for it. Based on the cost calculation, the library the increasing cost pressures that university libraries currently
manager can introduce a price differentiation between high experience, in particular related to ILL services.
service and low service ILL, with the first being more Hence, this paper bridges the library and cost accounting
expensive than the latter. However, the library manager argued literatures. First and most importantly, our study contributes to
that the library mission (again) opposes such a price strategy. the library literature by offering an innovative technique to
The Arenberg library wants to offer the best possible service to either reduce ILL resource costs or renegotiate ILL service
everyone and not only to those able and/or willing to pay for it. prices. Based on the case study evidence of the Arenberg
Nevertheless, there is one category of ‘‘customers’’ that does library, we conclude that TDABC can contribute to the pro-
qualify for price differentiation, namely companies requesting vision of better library services at lower costs. Yet, our paper
books or journal articles. Indeed, several large pharmaceutical also contributes to the cost accounting literature by testing the
and technology companies are frequent ILL users. Based on the usefulness of a new revised ABC technique in practice. In that
TDABC information of our analysis, the library manager could respect, our study is one of the first to offer clear evidence that
discuss the current service level and easily argue that he needs TDABC does accomplish the postulated benefits, when
to either raise the ILL price or lower the ILL service, in order to performed in practice.
cover the ILL expenses. Our time equations and cost Finally, we want to emphasize some interesting future
calculations support such a discussion with the necessary research opportunities on this topic. Obviously, the results of
relevant evidence, and form a benefit confirmed by the library our case study would be more informative when compared to a
manager. Finally, the library manager could consider a similar reference. Therefore, it would be interesting to perform a similar
discussion with the university management in order to request case study of TDABC for ILL in another (university) library.
and justify an increase in library funding. Based on the resulting data, a benchmark study could offer
interesting additional information to lower the costs or to
CONCLUSION increase the quality of the ILL service. Moreover, it looks very
In this paper, we have set up a time-driven activity-based promising to apply the TDABC approach to other library
costing (TDABC) system for the inter-library loan (ILL) services. In that respect, the Arenberg library manager suggested
service of the KULeuven Arenberg library. Based on cost performing a TDABC analysis on the activity ‘‘acquisition’’,
tables III and IV, borrowing costs appear to range from o4.23 which has similar characteristics (complex, time consuming and
(for the simplest article delivery) to o14.79 (for the most lots of variation) suggesting a high TDABC usefulness.
complex book loan), while incoming request costs range from
o2.12 (for the simplest article delivery) to o8.93 (for the most Acknowledgments: We wish to thank the KULeuven Aren-
complex article delivery). These results are comparable with berg library staff, and especially Ludo Holans (library
the cost rates found by the most recent ILL cost study of the manager) and Monique Kumps (ILL responsible) for their
Association of Research Libraries:14,15 borrowing unit costs for cooperation and contribution to this paper. In addition, we
user-initiated ILL ranging from $2.39 to $14.7 and user- gratefully acknowledge the assistance of Wannes Balieu and
initiated lending unit costs ranging from $3.27 to $12.06. Yet, Silvie De Roose in collecting the data. Finally, we also thank
they are slightly smaller than average costs (o12.56 for an anonymous reader for helpful comments that improved the
borrowing and o7.24 for lending) found in the Nordic13 and paper. Financial support by the Fund for Scientific Research,
significantly lower than the total average per unit cost of Flanders is gratefully acknowledged.
$41.20, found at the NIST Library.19 At the same time,
however, the KULeuven library’s ILL costs appear several NOTES AND REFERENCES
times bigger than the circulation cost around 1 dollar at
1. Paul Cropper & Roger Cook, ‘‘Activity-Based Costing in Uni-
OhioLINK17 or Prospector.18 Nevertheless, the most significant versities-Five Years On,’’ Public Money and Management 20
benefit of our TDABC analysis lies in the ability to (April-June 2000): 61 – 68.
disaggregate per-transaction costs, based on a thorough activity 2. Jennifer Ellis-Newman & Peter Robinson, ‘‘The Cost of Library
analysis that in the end visualizes the true cost of different Services: Activity-Based Costing in an Australian Academic
activities for different ILL requests. That way, the analysis also Library,’’ The Journal of Academic Librarianship 24 (September
(and perhaps most importantly) allows managerial recommen- 1998): 373 – 379.
dations concerning amelioration opportunities. For example, 3. Jennifer Ellis-Newman, ‘‘Activity-Based Costing in User Serv-
since TDABC makes it possible to distinguish the scanning of ices of an Academic Library,’’ Library Trends 51 (Winter 2003):
PDF files as a resource intensive activity, the library manager 333 – 348.
can use this information to consider the investment in a 4. Maurice Pendlebury & Nabih Algaber, ‘‘Accounting for the Cost
of Central Support Services in UK Universities: A Note,’’
professional scanner. Furthermore, the TDABC analysis shows Financial Accountability and Management 13 (August 1997):
that ‘‘searching’’ and ‘‘providing feedback’’ consume a large 281 – 288.
amount of resources. Based on these findings, potential 5. Andrew Goddard & Kean Ooi, ‘‘Activity-Based Costing and
ameliorations can be investigated, such as outsourcing the Central Overhead Cost Allocation in Universities: A Case
search activity to the user and improving the feedback process Study,’’ Public Money and Management 18 (July-September
by means of a FAQ portal. Hence, our case study not only 1998): 31 – 38.

September 2007 559

 6. Robert S. Kaplan & Steven R. Anderson, ‘‘Time-Driven Activity- 13. Pentti Vattulainen, Performance of Interlending in Nordic
Based Costing,’’ Harvard Business Review 82 (November 2004): Academic Libraries. Online. NORDINFO Board. Available:
131 – 138. http://inet.dpb.dpu.dk/nvbf/perform.pdf (August 22, 2003).
7. Danuta A. Nitecki & Patricia E. Renfro, ‘‘Borrow Direct: A Case 14. Mary E. Jackson, ‘‘Assessing ILL/DD Services Study: Initial
Study of Patron-Initiated Interlibrary Borrowing Service,’’ The Observations,’’ ARL Bimonthly Report 230/231 (October/Decem-
Journal of Academic Librarianship 30 (March 2004): 132 – 135. ber 2003): 21 – 22.
8. Julia C. Blixrud, Assessing Library Performance: New Measures, 15. Mary E. Jackson, ‘‘Assessing ILL/DD Services: New Cost-Effective
Methods, and Models. Online. Association of Research Libraries. Alternatives,’’ ARL Bimonthly Report 236 (October 2004): 9.
Available: http://www.iatul.org/conference/proceedings/vol13/ 16. David F. Kohl, ‘‘How the Virtual Library Transforms Interlibrary
papers/BLIXRUD_fulltext.pdf (2003). Loans—The OhioLINK Experience,’’ Interlending and Document
9. Joan E. Stein, ‘‘Measuring the Performance of ILL and Document Supply 26/2 (1998): 65 – 69.
Supply: 1986 to 1998,’’ Performance Management and Metrics 2/ 17. David F. Kohl & Tom Sanville, ‘‘More Bang for the Buck:
1 (2001): 11 – 72. Increasing the Effectiveness of Library Expenditures Through
10. Marilyn M. Roche, ARL/RLG Interlibrary Loan Cost Study, edited Cooperation,’’ Library Trends 54 (Winter 2006): 394 – 410.
by (Washington, DC: Association of Research Libraries, 1993) 18. Franca Rosen & Leanne Emm, ‘‘The Cost of Getting What
p. 73. Patrons Want: A Study in Colorado Resource Sharing,’’ Colorado
11. Mary E. Jackson, ‘‘Measuring the Performance of Interlibrary Libraries 29/3 (2003): 35 – 38.
Loan and Document Delivery Services,’’ ARL: A Bimonthly 19. Rosa Liu & Briget Wynne, ‘‘Analysis of Interlibrary Loan at the
Newsletter of Research Library Issues and Actions 195 (December National Institute of Standards and Technology Library: Lessons
1997): 1 – 3. Learned,’’ Interlending and Document Supply 34/3 (2006): 115–118.
12. Tom Ruthven & Susan Magnay, ‘‘Top Performing Interlending 20. Peter J. Lor, ‘‘Measurement of Suppliers’ Costs in a Large
Operations: Results of the Australian Benchmarking Study,’’ Decentralized Interlending System,’’ South African Journal of
Interlending and Document Supply 30/2 (2002): 73 – 79. Library and Information Science 60/3 (1992): 7 – 14.

560 The Journal of Academic Librarianship