1

FACULTEIT ECONOMIE
EN BEDRIJFSKUNDE

HOVENIERSBERG 24
B-9000 GENT
Tel. : 32 - (0)9 – 264.34.61
Fax. : 32 - (0)9 – 264.35.92

WORKING PAPER

Modeling Logistics Costs using Time-Driven ABC:

A Case in a Distribution Company

Werner Bruggeman1
Patricia Everaert2
Steven R. Anderson3
Yves Levant4

September 2005
2005/332

1
Department of Accounting & Corporate Finance, Faculty of Economics and Business Administration, Ghent
University, Belgium, e-mail: [email protected]
2
Department of Accounting & Corporate Finance, Faculty of Economics and Business Administration, Ghent
University, Belgium, e-mail: [email protected]
3
Acorn Systems, Inc., Houston, TX USA, e-mail: [email protected]
4
Université des Sciences et Technologies de Lille and Lille Graduate School of Management, France, e-mail:
[email protected]

Corresponding author: Patricia Everaert, Faculty of Economics and Business Administration, Ghent University,
Kuiperskaai 55/E, 9000 Gent, Belgium, phone: 32 9 264 35 00, e-mail: [email protected]

D/2005/7012/50
 2

Abstract

Purpose: The purpose of this paper is to study whether time-driven Activity-based Costing
(ABC) is more appropriate to capture the complexity of logistics transactions than traditional
ABC.
Methodology: This paper explores the method of time-driven ABC and presents the results of
a case study at a distribution company.
Findings: The case shows that a traditional ABC was not appropriate. Almost 60% of the
activities contained more than one subtask, with different time drivers. Ignoring this
complexity resulted in a misallocation of 54% of costs. Splitting-up all the subtasks resulted
in a traditional ABC model that was too complex. Contrary, the time-driven ABC model
captured the full complexity of the logistics transactions through time equations that employ
multiple time drivers.
Research implications: This paper formulates the underlying mathematical model of time-
driven ABC, leading to a better understanding of the time equations. Different types of
complexities in logistics were identified in the case.
Practical implications: The study explains the technique of time-driven ABC and provides
some real company examples of time equations in logistics.
Value of the paper: This is one of the first papers on time-driven ABC in a European context.
The mathematical representation allows to a better understanding of this new costing
technique and provides a new methodology to model complexity of logistics activities.

Keywords: Logistics, Time-driven Activity-based Costing, time equations, mathematical

model, complexity

Classification: Conceptual Paper and Case Study

 3

1 Introduction

While supply chains have existed ever since businesses have been organized to bring products

and services to customers, the notion of their competitive advantage, and consequently supply

chain management (SCM), is a relatively recent thinking in management literature (Chen &

Paulraj, 2004). In the relationships between suppliers and their retail customers cost

knowledge is identified as essential for two main reasons. First, it can establish a competitive

advantage by focusing on the most profitable customers (Foster et al., 1996). Second,

managers must know the unit cost of products as well as of customer service levels, to use

during negotiations with other members of the supply chain in order to realize fair exchanges

and equitable partnerships (Norek & Pohlen, 2001; Lin et al., 2001; Themido et al., 2000).

Many suppliers contend there is an increasing amount of pressure placed on them to perform

an increasing number of functions previously performed by their retail customer (such as

special packaging requirements) (Norek & Pohlen, 2001). Without knowing the cost-to-

serve, suppliers have little evidence to demonstrate how shifting functions in the supply chain

effects their profitability and long-term survival (Cokins, 1999).

In reporting profitability of orders and customers, traditional cost systems treat logistics costs

as overhead costs and allocate these costs to products/customers on the basis of volume-

drivers such as direct labor hours, units or turnover (Cooper, 1988a). Using volume-based

allocation rules for logistic costs might lead to distortions in cost and profitability. The ever-

increasing range of products and services offered and the wide variety of channels and

customers make it difficult for logistics costs to be modeled by time-drivers alone (Pohlen &

La Londe, 1994). Since more than two decades, many companies have successfully

developed activity-based cost (ABC) systems for logistics activities (Pohlen & La Londe,
 4

1994; Lin et al., 2001; Themido et al., 2000). ABC directly links the costs of performing

organizational activities to the products, customers or distribution channels. These links are

usually unknown in advance and have to be identified at the level of the individual activities

in the different overhead departments (Cooper & Kaplan, 1998). In complex and dynamic

environments many companies abandoned activity-based costing because it often failed to

capture the complexity of actual operations, it took too long to implement and was too

expensive to build and maintain (Kaplan and Anderson, 2004). To solve these problems

Kaplan and Anderson introduced the concept of time-driven activity-based costing. Time-

driven ABC provides the ability to identify and report complex and specialized transactions in

a simple way by using time equations that can employ multiple drivers.

The concept of time-driven ABC is relatively new and unexplored in current research (2004).

Because logistics activities are expected to be complex, the purpose of this paper is to study

whether time-driven ABC is more appropriate to capture the complexity of logistics

transactions than traditional ABC. We collected data from a case study at a distribution

company. This company deals with many different products and types of customers, asking

for a variety of service in an intensely competitive environment. During the case study the

following specific research questions were considered. (1) Why does this company need a

time-driven ABC system? (2) What kind of activity complexity needs to be captured to

provide accurate cost and profitability information about orders and customers? (3) Why was

traditional ABC not able to fully capture the identified complexities? (4) How could the

identified complexities be incorporated in the time-driven ABC model?

 5

There are several goals of this paper. The first objective of this paper is to review the

shortcomings of traditional ABC systems with the opportunities of time-driven ABC, both

from a conceptual and empirical point of view. The second objective lies in a clear definition

of the underlying mathematical model of the technique of time-driven ABC and the concept

of time equations. By utilizing multiple time drivers and interaction terms, we witnessed

time-driven ABC’s unique way to describe the complexity of logistics activities. Third, this is

the first study (as far as we know), describing and exploring a time-driven ABC model at an

European company. Finally, this study shows how innovative cost management techniques

can be successfully applied in supply chain management.

The paper is organized as follows. First, we address the technique of traditional ABC and the

difficulties encountered in applying ABC. Then, we discuss the technique of time-driven

ABC. Next we develop a mathematical model underlying time-driven ABC to explore

whether time-driven ABC can solve the shortcomings of a traditional ABC system in complex

environments. Further on, we describe the case study and analyze the time equations used in

this distribution company. Finally, we give a summary of the type of complexities we could

identify in the logistics cost model of this company.

 6

2 Activity-based Costing and its reported Shortcomings

2.1 Technique of ABC

Activity-Based Costing is a costing method that first assigns overhead costs to activities and

then to products, orders or customers, based on how much each of these cost objects uses the

individual activities (Cooper, 1988a, 1988b, 1989a, 1989b; Cooper & Kaplan, 1988; Kaplan

& Cooper, 1998). ABC typically involves the following steps, as shown in Figure 1:

Step 1: Identify the different overhead activities.

Step 2: Assign the overhead costs to the different activities by a resource driver.

Step 3: Identify the activity driver for each activity.

Step 4: Determine the activity driver rate by dividing the total activity costs by the

normal volume of the activity driver.

Step 5: Multiply the activity driver rate by the activity driver consumption to trace

costs to cost objects.

Insert Figure 1 about here

For example, consider the activity “Sales Order Processing”. Assume that for this activity the

total cost (payroll, depreciation and other supplies) is equal to €50.000. When the number of

orders is the activity driver and 1.000 orders can be processed, ABC will come up with an

activity driver rate of 50.000/1.000 or €50 per order. In calculating order profitability each

order will be assigned €50 per order for sales order processing costs, independent of the

number of units that the customer is ordering.

 7

Previous research illustrated successful implementations of ABC in logistics (Pohlen & La

Londe 1994; Cokins, 1999; Themido et al. 2000; Lin et al. 2001). In the general ABC

literature, many studies have focused on measuring the success of ABC and discovering its

determinants (see Foster & Swenson 1997 for a review). We will focus in this literature

review on the difficulties encountered when applying ABC. Afterwards, when we discuss the

technique of time-driven ABC, we will evaluate whether these difficulties still apply under

time-driven ABC.

2.2 Difficulties encountered in applying ABC

Measurement errors leading to less accurate cost information

ABC systems are set up to reduce specification errors. These errors arise when a volume-

based allocation is used, while in reality costs are often driven by non-volume related

activities. ABC systems are also intended to reduce aggregation errors, since more accuracy

occurs as more activity cost centers and cost drivers are used to assign costs to products

(Datar & Gupta, 1994). But the refinements of cost systems to reduce these errors might

come at the cost of higher measurement errors (Datar & Gupta, 1994). Information on the

exact cost or the nature of resource usage at less aggregate cost pools are often difficult to

obtain (Foster & Gupta, 1990; Lin et al., 2001). Because of these measurement problems,

companies might shift to a less detailed ABC system.

Measurement of costs or units of activity drivers are further complicated when variables to be

measured are not supported by well-defined measurement guidelines or measurement

techniques. For example, consider the activity “Order Processing”. Computing total costs

incurred in order processing requires estimates of the percentage of time spent by various staff

functions such as sales representatives, credit controllers and planners. The firm typically
 8

estimates the time spent on the basis of questionnaires and interviews (Cooper et al. 1992)

and these estimates are subject to measurement error (Datar and Gupta, 1994).

Enterprise-wide ABC models are becoming too complex

Anderson et al. (2002) found more complex ABC models for companies facing a great deal of

competitive pressure. In a highly competitive environment, the need to focus on developing

accurate product costs increases, because plant managers need more data on the cost of

operations and opportunities for improvement. More accurate product costs are obtained in

ABC as overhead support activities are broken down into finer components (Kaplan &

Cooper, 1998). Also Kaplan & Anderson (2004) argue that when the activities become more

advanced, ABC requires that activities have to be split into smaller activities leading to an

inflation of the number of activities. For example when the costs of order processing not only

depend on the number of orders but also on the type of customer, accurate costing should use

different order processing cost rates for every type of customer. An overproliferation of

activities and activity drivers are frequently found for staff functions (such as logistics and

distribution), since these employees face less repetitive tasks than on the shop floor (Innes et

al.,1992; Armstrong, 2002).

Kaplan and Anderson (2004) argue that as the activity dictionary expands either to reflect

more accuracy and detail or to expand the scope of the model to the entire enterprise, the

demands on the computer model used to store and process the data escalate dramatically. For

example, a company using activities in its enterprise ABC model, and applying the costs in

these 150 activities to 600.000 cost objects (SKUs and customers) and running the model

monthly for two years requires data estimates, calculations and storage for more than 2 billion

items (Kaplan & Anderson, 2004, 132). To reduce the difficulties of operating an enterprise-

wide ABC model, companies often build separate ABC models for each of their sites or start
 9

with one product group, one time period and one type of distribution channel (Themido et al.,

2000). Trying to coordinate cost estimates for products/customers traversing multiple ABC

models then becomes almost impossible.

Time-consuming to build a complex ABC model

Many ABC models languish when they are too complex, which in turn leads to long

development time (Anderson et al., 2002). Survey research in different countries show that

the amount of work is experienced as one of the biggest problems for the team designing the

ABC system as well as for the accountants implementing it (Cobb et al., 1994; Pohlen & La

Londe, 1994). The analysis of activities involves many interviews, typically requiring about

three people working full-time for between four and six months (Cooper, 1990). Very often

activities are crossing departmental boundaries and this means coordination of information

from many interviews to determine the major activities, again very time-consuming (Cobb et

al., 1994).

Difficult to update a complex ABC model

Armstrong (2002) argues that since an ABC model is extensive in staff time to install, it will

be even more expensive to update. Kaplan and Anderson (2004) observe that in dynamic

environments, where activities, processes, products and customers frequently change, ABC

might lead to a high cost of continually updating the model, out-of-date activity driver rates,

and inaccurate estimates of product, process and customer costs. Consider the following

example in the activity “ship order to customer”, taken from Kaplan and Anderson (2004).

Rather than assuming a constant cost per order shipped, a company might wish to recognize

the cost differences when an order is shipped in a full truck, in a less than truckload shipment
 10

or using overnight express. In addition, the ship order might be entered either manually or

electronically, and it might require either a standard or an expedited transaction. To allow for

the significant variation in resources required by the different shipping arrangements, new

activities must be added to the model, thereby expanding its complexity.

3 Time-Driven Activity-based Costing

The concept of time-driven ABC was originally developed in 1997 by Steven Anderson and

practiced through his company Acorn Systems, Inc. In 2001, Anderson teamed up with

Robert Kaplan of Harvard Business School to perfect the approach. This was consummated

in several white papers that the two wrote starting later that year. This new approach requires

only two parameters to estimate: the unit cost of the supplying resources and the time required

to perform an activity by this resource group. The time-driven approach of ABC consists of

the following six steps (Kaplan & Anderson, 2004):

1. Identify the various groups of resources that perform activities.

2. Estimate the cost of each group of resource.

3. Estimate the practical time capacity of each group of resource (i.e. available working

hours).

4. Calculate the unit cost of each group of resource by dividing the total cost of the

resource group by the practical capacity.

5. Determine the required time for each event of an activity based on different time

drivers.

6. Multiply the unit cost by the time required to trace costs to cost objects.
 11

Remark that the breakthrough of time-driven ABC lies in the time estimates (see step 5

above). Kaplan & Cooper (1998, 294) earlier suggested in a traditional ABC system to use

the capacity of the resources supplied for assigning resource expenses to activities (see step 3

and 4 above). New in time-driven ABC is that the time required for performing the activity is

now estimated for each event based on different characteristics, the so-called time-drivers.

The basic methodology of time-driven ABC is shown in Figure 2.

Insert Figure 2 about here

Again consider the example of the sales order processing activity. Assume a total resource

cost of €57.600 (payroll, depreciation, other supplies) per week related to a practical time

capacity of 5.760 minutes (80% of the theoretical capacity of 40 hours per week for 3

employees). So, the cost per minute for this group of resources is 10 euro. The time required

to process a standard order is estimated to be 3 minutes. Order processing for new customers

requires a subtask of registration, taking an additional 15 minutes. So the event of order

processing for a new customer takes 18 minutes, while the event of order processing for an

existing customer takes 3 minutes. Hence, in the time-driven approach, the cost per order

equals €30 for existing customers and €180 for new customers. By no longer using a

transaction driver (number of orders) but the time required to perform the order processing

activity, the cost per order can be made fully situation dependent, without rebuilding the

whole model. Based on the order processing characteristics, event 1 of order processing

might take 3 minutes because it is an existing customer, event 2 might consume 18 minutes

because it is a new customer, while event 3 might require again 3 minutes, etc. Hence, the

time-driven ABC uses duration drivers (such as set-up hours, material handling time, order
 12

processing time) instead of transaction drivers (such as the number of set-ups, the number of

material moves, the number of orders). The reason is that in complex environments a

particular activity does not always consume the same quantity of resources in every situation.

Rather than defining a separate activity for every possible combination of order processing

characteristics, the time-driven approach estimates the resource demand by a time equation

(Kaplan & Anderson, 2004). The time equation for the simple example above is:

Order processing time per order = 3 + 15if new customer

4 Our Mathematical Model of Time-driven ABC

As mentioned above, the cost of an activity is generated by the time required for the activity

multiplied by the cost per time unit. Remember from the example in the previous paragraph

that the time required for the order processing activity is calculated for each specific event of

order processing, depending on the characteristics of that specific event k. We represent this

in the following mathematical way:

depend on the characteristics of that specific order.

Cost of an individual event k of Activity J = t j, k . c i

With ci: the cost per time unit (minute) of resource pool i
tj, k: the time consumed by event k of activity j

The total cost over all events of all activities is then calculated by summing up all activity

costs. The total cost of a cost object (e.g. customer, order, product) can be calculated by:

n m l
Total Cost of a Cost Object = ∑ ∑∑ t j, k . c i
i =1 j =1 k =1
 13

With ci: the cost per time unit (minute) of resource pool i
tj, k: the time consumed by event k of activity j
n = the number of resource pools
m = the number of activities
l = the number of times activity j is performed (or the number of events of a particular
activity j)

By using time equations, the time consumed by the event of an activity (tj,k) can be expressed

in function of different characteristics, the so-called time drivers. The following general time

equation describes the time needed for an event k of activity j, with p possible time drivers X.

β0 represents the constant time, independent of the characteristics of the activity. Β1 indicates

the increase in time for one unit increase in X1 (when X2, …, Xp are held constant).

t j, k = β 0 + β1.Χ1 + β 2 .Χ 2 + β3 .Χ 3 + ... + β p .Χ p

with: tj,k = time required to perform event k of activity j

β0 = constant amount of time for activity j, independent of the characteristics of event
k
βl = time consumption for one unit of time driver 1
X1 = time driver l, X2= time driver 2, … , Xp = time driver p,
p = the number of time drivers that determine the time needed to perform activity j

5 Exploring the Time Equations in Time-Driven ABC

5.1 Different type of Time Drivers

Time drivers are essential in time-driven ABC. As defined above, time drivers are variables

(characteristics) that determine the time needed to perform an activity. They can take the

form of a continuous, discrete or indicator variable. We took the following examples of the

case (as described later) to illustrate the difference.

• Continuous variables: e.g., the weight of a pallet, distance in kilometers.

 14

• Discrete variables: e.g., the number of orders, the number of order lines, the number of

credit checks.

• Indicator (or dummy or 0-1) variables: e.g., the type of customer (old versus new), the

type of order (normal versus rush), the order reception characteristic (EDI versus fax).

The additional time is then added up in the equation when the characteristic is present

(X3 = 1 for an order received by fax) and no additional time is added up when the

characteristic is not present (X3 = 0 for an EDI received order).

5.2 Multiple time drivers

The major advantage of time-driven ABC is that multiple drivers can be taken into

consideration to define the cost of an activity. In traditional ABC only one activity driver can

be considered for each activity. If many activity drivers are necessary for accurate costing,

different activities are needed in ABC (e.g. “order line entry” driven by the number of order

lines, “new customer registration” driven by the number of new customers). In time-driven

ABC the number of time drivers (or terms in the time equation) is not limited as long as the

employees performing the tasks are belonging to the same resource pool (i.e. consuming the

same ci).

5.3 Interaction of time drivers

Also time equations might take into account interactions between drivers (e.g., the time

required for registering a new customer might differ whether the customer is on the phone or

whether the data are coming from a sales representative). The general time equation

including the main effects and the two-way interactions for two time drivers can be written in

the following general time equation:

 15

t j, k = β 0 + β1.Χ1 + β 2 .Χ 2 + β3 .Χ1.Χ 2

In the following sections, we provide some examples, taken from our experience with the

company from the case study (as will be discussed on page 19).

5.4 Example of discrete and indicator time drivers

Let’s elaborate the introductory example. The order processing might now depend on three

time drivers: the type of customer (new or existing customer), the number of line items and

the order type (normal or rush order). Order entry of the basic order information involves 3

minutes, each line item requires 2 minutes data input, input of new customer data takes 15

minutes and 10 additional minutes are required for expediting when the order is a rush order.

We have now one discrete time driver (i.e. the number of order lines, represented by X1) and

two indicator time drivers, i.e. X2 (new versus existing customer) and X3 (normal versus rush

order). The indicator variable X2 takes the value of 1 when the customer is new and 0 for an

existing customer. The same applies to indicator variable X3: the additional time is only

required for a rush order, assuming a “1” for a rush order, and a “0” for a normal order. The

estimated order processing time is calculated by the following time equation:

Order processing time per order = 3 + 2 * X1 + 15 * X2+ 10 * X3 [1]

with: X1 = number of order lines

X2 = New customer (1) versus existing customer (0)
X3 = Rush order (1) versus normal order (0)

The order processing time of a rush order with 5 order lines for a new customer will generate

the following time equation:

Order processing time tk = 3 + 2 * 5 + 15 * 1+ 10 * 1 = 38 minutes

 16

5.5 Example of a two-way interaction in time drivers

Let’s elaborate the example with one more characteristic. The order processing time now also

depends on the type of customer. Because of the advanced technical specifications present in

the order line descriptions, the time per order line for customer XYZ does not take 2 minutes

but 10 minutes (so 8 more minutes than the base time per order line). In this case also the

type of customer should be included in the time equation as an indicator variable. In this

example the type of customer affects only the time per order line and hence it should be

included in the time equation as an interaction term, with a time estimate of 8 minutes. If

customer XYZ places the order, activity driver X4 gets a value of 1. Remark that in this

example there is no main effect of customer XYZ, since the time estimate for X4 is zero.

Order processing time per order = 3 + 2 * X1 + 8 * X1* X4 + 15 * X2+ 10 * X3+ 0 * X4

with: X1 = number of order lines

X2 = New customer (1) versus existing customer (0)
X3 = Rush order (1) versus normal order (0)
X4 = Customer XYZ (1) versus all other customers (0)

Using this information the time consumption of a normal order with 5 order lines placed by

existing customer XYZ can be calculated as:

Order processing time tk = 3 + 2 * 5 + 8 * 5* 1 + 15 * 0+ 10 * 0 + 0 * 1 = 53 minutes

5.6 Example of a three-way interaction in time drivers

Assume that in the previous example the additional time to process technically sophisticated

order line descriptions depend on the technical knowledge of the person processing the order.

Suppose the additional time is 8 minutes when the order processing is performed by a sales

administrative person and only 3 minutes when a technical person does the work. Also

assume that the cost of both resources (administrative and technicians) is the same. In this
 17

case order lines of an order placed by customer XYZ still generates 8 more minutes per order

line, except when the order is processed by a technical person. Such a person spends 5

minutes less per order line than an administrative person. To include this time effect into the

time equation a fifth variable has to be defined: the type of order intake person, X5, being an

indicator time driver. In this modeling problem, we now face a three-way interaction between

the number of order lines (a discrete variable), the type of customer and the order intake

person (both indicator variables). This is the general time equation:

Order processing time per order = 3 + 2 * X1 + 15 * X2+ 10 * X3+ 3 * X1 * X4

+ 5 * X1 * X4 * X5

with: X1 = number of order lines

X2 = New customer (1) versus existing customer (0)
X3 = Rush order (1) versus normal order (0)
X4 = Customer XYZ (1) versus all other customers (0)
X5 = Order intake by sales (1) versus order intake by technical department (0)

The calculation of the time to process a normal order with 4 order lines, placed by customer

XYZ and processed by a technical person involves 23 minutes, as indicated in the following

equation:

Order processing time tk = 3 + 2 * 4 + 15 * 0 + 10 * 0 + 3 * 4 * 1 + 5 * 4 * 1 * 0 = 23 minutes

5.7 Example of changing time drivers

We return to the example with three time drivers (see [1]). Suppose the finance department

urges sales administration to do a credit check, from now on, before accepting and processing

a customer order. Because the credit check is going to be executed by the same resources, we

don’t need to change the cost system and add a new activity. The requirement of doing a

credit check each time an order comes in, will only increase the time per order of the order
 18

processing activity. Assume that a credit check will take 10 minutes, the constant term will

increase from 3 to 13 minutes and the time equation becomes now:

Order processing time per order = 13 + 2 * X1 + 15 * X2+ 10 * X3

with: X1 = number of order lines

X2 = New customer (1) versus existing customer (0)
X3 = Rush order (1) versus normal order (0)

5.8 Conclusion: Can Time-driven ABC solve the ABC shortcomings?

These examples illustrate that time-driven ABC can solve some of the earlier mentioned

difficulties encountered with traditional ABC. (1) The measurement of time in minutes (or

hours) on a specific activity as a function of different characteristics (time drivers) can reduce

measurement errors. Staff people can be more specific on estimating the time of a broad

activity (with all its variants) than to estimate the total time spent for a whole year on each of

the variants of the activities. (2) The use of time equations allows companies to design time-

driven enterprise wide systems with a limited number of activities. As Anderson & Kaplan

have shown through the inclusion of multiple time drivers, complex activities can be modeled

without expanding the number of activities. Hence, time-driven ABC provides apparently

many opportunities to design ABC models in environments with complex activities, such as

in logistics and distribution companies, hospitals, and servicing companies in general. The

second part of this paper will further explore these opportunities for the logistics activities in a

distribution company. (3) The time equations also allow easily updating the cost system

when products or service offerings change or when production and service processes are

redesigned. This characteristic makes time-driven ABC suitable for fast changing

environments and makes it a candidate for corporate agility. Furthermore, the unit cost per

resource pool can be easily updated when wages, ICT usage or tooling changes within the
 19

resource pool, without recalculating the whole model. (4) Finally, based on their experience

with more than 100 clients, Kaplan & Anderson (2004) believe that time-driven ABC systems

can be estimated and installed quickly.

Future research on time-driven ABC applications is needed to confirm whether the above-

mentioned opportunities of time-driven ABC will counteract the reported problems with ABC

in practice. The technique of time-driven ABC looks promising, but more case studies are

needed to reinforce our knowledge on this innovative topic (Ryan et al., 2003). In this study,

we did a case study to further explore the opportunities of time-driven ABC to design a cost

model for complex activities.

6 Case Study

6.1 Methodology

In order to explore real-world implementation of a time-driven ABC, we conducted case

study research at a distribution company, during May-June 2004. A case study is the most

appropriate research method to explore the reasons for a particular accounting practice

(Scapens, 1990; Otley & Berry, 1994). As such, this case study represents a preliminary

investigation of an innovative practice, which is intended to generate ideas for testing at a

later stage.

Data was collected using various sources. Interviews were conducted with the managing

director, the controller and the consultant who had installed the time-driven ABC system.

Since the purpose of our study was to learn from practice, we conducted open-ended

interviews. Rather than assuming that the existing literature had covered all aspects of time-
 20

driven ABC, we tried to uncover new concerns as much as possible. Therefore, the direction

that every interview took was unique. Other documents such as annual reports, internal

reports, promotional brochures, a video and power point presentations were studied.

Complete access was given to the activity database and the time equations.

The authors approached this company because of a business contact between one of the

authors and the consultant. In a sense, this site selection was not ad random, although it

helped tremendously to overcome one of the biggest practical difficulties of conducting case

study research, i.e. gaining access to field sites (Baxter & Gua, 1998). This company was

attractive on a priori, objective grounds, and would have been a top candidate in a purposive

sampling approach, because it implemented the time-driven approach company-wide and was

already using the time-driven cost information for almost one year at the time of the

interviews. We admit that the method of access might be a source of bias and a threat to

internal and external validity of the qualitative case results (e.g. need for time-driven ABC or

reported benefits), while we do not expect any bias towards the more quantitative case results

(e.g. time equations or complexity of the ABC model).

The goal of this study is to explore the time-driven approach of ABC in practice. In particular

we wanted to discover why the company was using time-driven ABC to contrast with the

academic literature’s assumptions regarding the need for time-driven ABC. Secondly, we

want to explore whether time-driven ABC can capture the complexity of the logistics

operations. Third, we want to find out why traditional ABC was not able to capture the

complexity in the logistics operations. Finally, we wanted to find out how this company

translated the complexity of the activities into the time equations.

 21

6.2 Research Site

Facts

Sanac is a distributor of plant-care products in Belgium with total sales of 62 million euro, an

own transportation fleet of 25 trucks and a warehouse of 22 500 pallets. Sanac neither

produces nor sells any of its products under its own name. It only retails well-known brand

products to four different types of customer groups: farmers, growers, public sector

companies & landscapers, and large and small consumer shops. Sanac has about 7.000

customers, 7.000 products, and 298.000 order lines per year. Sanac employs 129 people; 40

of whom are employed as sales technicians and 57 in the logistics department. Sanac is in

continuous expansion and in the last ten years its sales has increased by 10% per year.

Competitive Strategy

There are three very important elements of the competitive strategy. The first one is advice.

Sanac sells not only items related to planting and growing (especially the plant-care products)

but also indirectly, advice. They consider it as essential to provide technical support for the

sale of these products. Technical skills and the services and advice from technicians on the

products together with solutions to technical problems, lead to sales as part of a working

partnership that in the long-term ensures a reliable and trusting relationship with its

professional clients. Advice for the small and large consumer shops are realized by offering a

display, as shops do not usually have either the structure or the time to deal with technical

products or to give advice to customers. Sanac, in fact, runs a sales area for one store

ensuring its profitability without any or with very little investment in time or training for the

supermarket employees. Profitability per square meter is very important for supermarkets and

hypermarkets. The second element of the competitive strategy is marketing and knowledge of

the market. The markets where Sanac is present have a seasonal-based activity and are

subject to trends. Therefore, feedback from the technicians and sales people as well as from
 22

the truck drivers is essential as it is a precious source of information and helps to maintain

contact with clients in order not to have any unsold items. Third, Sanac is also an expert in

logistics. Logistics are important, as the company is either having to deal with professionals

within supermarkets and hypermarkets for whom any delay means a loss of sales or with

professionals for whom speed of delivery is an essential asset in order not to prejudice their

crop.

Market Evolution

In the nineties the company was very sales driven, without any insight into the profitability of

the different customers. At the end of the nineties competition increased, profit margins

became under pressure and customers asked more services. The company also experienced a

faster changing market situation, strong seasonal fluctuation in its activities and a growing

complexity and diversity of its products and its customers. These changes in the environment

forced the company to change from a growth strategy to a profitability enhancing strategy.

Marketing policy was focused on the consumer shops, which was a new segment. But

nobody knew the true cost of this new activity or the resources invested in it (the logistics,

terms of payment). Despite an important increase in sales (from 37 to 62 million euros in four

years) and stable gross profit margins, a decrease in the overall result was noticed. In order to

solve what was considered a paradox, Sanac tried to calculate profitability by sector. The

existing cost accounting system only enabled a contribution margin (gross margin minus

logistics and commercial costs) to be calculated. There remained however unallocated

indirect overheads (2.5 million euros in 2000) and these costs, considered in principle as

fixed, were increasing each year. Even with incomplete data, Sanac saw that the consumer

segment was making a big loss, that the agriculture was profitable and that the horticultural

sector was loss making. In fact, the sector where there was most growth and where Sanac
 23

wanted to re-orientate itself was the least profitable. Sanac was changing its business but this

new business was not understood financially.

6.3 The Need for a Time-Driven ABC model at Sanac

The need for a more accurate cost system came from reduced profitability. Sanac’s aim was

to recover its profitability, to shift its focus from growth (a focus forcing the company to have

a wide range of products with large stocks and to deliver any order to any client in the shortest

possible time and in whatever quantity was needed at a single price) to profitability. The firm

wanted to know what its costs were per product, per customer, per delivery, per order line and

per supplier (see Figure 3). All this effort was meant to lead to improvement in services and

to privilege clients who generate value, that is to say margins and profit for Sanac, even if this

meant discarding certain clients and/or products or even markets. For this to occur, there had

to be a reliable system for measuring costs and performance before there could be any

changes in strategy.

Insert Figure 3 about here.

Sanac started with a traditional ABC, but soon realized that this ABC cost system would not

work out. Several reasons were mentioned. (1) The CEO wanted monthly profitability data

at the level of business unit, products, orders and customers. This required that the model be

updated monthly with actual data. This was not possible without hiring 10 controllers. (2)

The rapidly growing market of consumer shops required Sanac to take immediate actions on

customer’s requests. These clients are very demanding in terms of price and resources. In

such a dynamic environment the maintainability of the cost model was very important. Again
 24

without 10 controllers, keeping the ABC model up to data was impossible. (3) Strong

seasonal trends in the sales made it difficult to define the “normal capacity” of the activity

cost drivers (80% of the sales are achieved in four months with an idle capacity in winter).

During winter periods the company cannot charge all unused capacity to that one customer

that is coming in November. Working with averages was not working out in this company,

because the CEO wanted to know each month what departments were having overcapacity

and what departments were running under capacity. (4) High diversity of customers and

products required too many activities to keep track off. Sanac has 7,000 clients, a portfolio of

7,000 products in stock and 20,000 in catalogue. It receives 298,000 orders a year and issues

69,000 invoices. But more important, clients have important and varied demands in services

and in logistics. This wide diversity in the consumption of resources makes it difficult to

trace and analyze costs. For instance, there is a wide difference in unit costs according to the

products (packaging, type of product) and the clients (delivery terms, advice, terms of

payment). A small farmer does not generate the same gross profit margins or the same costs

as the DIY store of a large chain. A traditional ABC system required here too many

activities. (5) Last, but not least, the strong diversity in resource consumption by the various

cost objects could not be captured by a traditional ABC model. During the activity analysis

preparing for the implementation of a traditional ABC, they experienced that many logistics

activities were too complex and that cost rates, such as the cost per drop, the cost per order

etc. were too average and did not mean very much to the people interpreting the profitability

reports. The examples in the next paragraph illustrate the kind of complexities this company

was dealing with in its logistic activities.

 25

6.4 Time Equations of some Complex Activities

Supplier-related activity: Reception of incoming goods

Interview data reveal that reception of incoming goods is a process in itself, composed by

many subtasks, each with different time drivers. The interviewee described the task as

follows: “When the truck arrives the driver is assigned a dock (1 minute). Next the

warehouse employee enters the delivery voucher into the computer. Input of the general order

information takes 1 minute and input of each purchase order line takes 10 seconds per

purchase order line. Unloading of pallets takes 2,5 minute per pallet. Each order delivery

requires also 1 minute of administration (communicating information about reusable

containers or signalling of specific problems). When reusable containers are returned to the

supplier employees need to spend 10 minutes per pallet for sorting, counting and blistering.

Sometimes pallets have to be returned to the supplier. In this case pallets are transported in

staples of 15 from the warehouse and loaded on the truck. This operation needs 3 minutes per

staple of 15 pallets. When the truck is unloaded the delivered goods need to be checked,

prepared and stored in the warehouse. This operation takes 1 minute per homogeneous pallet

and 15 minutes per heterogeneous pallet. Finally a warehouse employee signs off the voucher

(1 minute).”

Analysis of these data shows that this activity is not homogeneous and can be subdivided in

different tasks, each having its own specific time driver and time consumption per unit time

driver (see Table 1).

Insert Table 1 here.

 26

The activity reception of incoming goods is not solely driven by the number of purchase

orders as would be used in a traditional ABC model. The activity can be considered as a

“multiple driver” activity, because multiple time drivers influence the activity time. The

following time equation is capturing the time drivers5:

Reception time per delivery = 4 + 0,17 * X1 + 2,5 * X2+ 10 * X3+ 3 * X4/15 + 1 * X5

+ 15 * X6
With: X1 = number of order lines
X2 = number of pallets received
X3 = Number of pallets to return with reusable containers
X4 = Number of pallets to return
X5 = Number of homogeneous pallets received
X6 = Number of heterogeneous pallets received

Using the time equation we can calculate the cost per delivery in different situations. For

instance a reception of incoming goods with 10 order lines, 10 (homogeneous) pallets, no

reusable containers or pallets to be returned requires 41 minutes. A reception of incoming

goods with 10 order lines, 10 pallets, 5 pallets with reusable containers and 2 staples of 15

(homogeneous) pallets to be returned ask for 97 minutes. The examples show that activity

times per order might significantly differ from each other.

Customer-related activities: drop off with the customer

In the traditional ABC model the identified activity driver was the number of drops.

Traditional ABC could only come up with an average cost per drop. But in reality the cost

per drop depends on many variables. The following interview data picture the complex reality

5
In the Acorn time-driven ABC software, this time equation was written as follows: 1 + 1 + 0.17 * [Line Items]
+ 2.5 * [ReceivedPallets] + 1 + 10 * [PalletsReusableContainers] + 3 * [EmptyPallets] / 15 + 1 *
[HomogeneousPallets] + 15 * [HeterogeneousPallets] + 1
 27

of the dropping activity: “Farmer routing: An order dropped at a farmer’s premises normally

takes 5 minutes per order, except when it is a first delivery. In this case the drop takes 10

minutes. When the truck driver also has to take back returned goods, he needs an additional 5

minutes to check the goods and to finish the necessary administration. When reusable

containers have to be taken back the truck driver needs 10 minutes to check the labels.

Delivery under cash payment requires an extra 20 minutes per drop delivered to farmers and

growers, and 5 minutes for all other customers. Non-farmer routing: When the customer is

not a farmer the time per drop is 2 minutes per pallet when the goods are delivered on pallets,

and 1 minute per pack when delivery is in packs. With some customers there are important

waiting times. In garden centres the waiting time is 15 minutes, in large distribution

companies the waiting time is 30 minutes per drop. However when the delivery goes to

Makro or Leroy Merlin the driver has to wait 60 minutes. Non-farmer customers might also

have goods to take back. When an appointment has been made, the goods are ready to load

and have been assigned a return code. In this case loading of returned goods only takes 3

minutes. When no appointment has been made the driver has to call Sanac to ask a return

code. This inquiry takes 30 minutes. In case of non-farmer customers loading and checking

empty pallets takes 2 minutes. Cash payment for non-farmers requires 10 minutes.”

The interview data show that the time per drop depends on the type of customer (farmer or

non-farmer), the type of non-farmer customer (large distribution company or not, garden

center or not, Makro/Leroy Merlin or not), the type of delivery (first delivery or not), the

availability of a return code (yes or no), the need to take back goods (yes or no), the need to

take back reusable containers (yes or no), the mode of payment (cash or non-cash), the

number of pallets and the number of packs, as shown in Table 2.

 28

Insert Table 2 here.

The time drivers affect the time of the drop-off activity in different ways. Some factors

determine an additional subtask during the delivery process (e.g. first time delivery needs

more time to talk with the farmer), some factors determine the type of time driver (e.g. a

farmer drop off takes 5 minutes per drop, a non-farmer drop off depends on the number of

pallets and the number of packs), and some factors determine the time per unit time driver

(e.g. return of goods ask 3 minutes for a non-farmer and 5 minutes for a farmer). The

example also illustrates how the time spent in a certain activity might depend on the event of

other activities: the subtask “taking back returned goods” takes more time when no

appointment has been made. Using time equations in time-driven ABC allows for modeling

interdependencies between activities. The following time equation was set up for this

logistics activity.

Delivery time per drop6 = 5 * X1+ 5 * X1. X2 + 5 * X1 X3+ 10 * X1 X4 + 5 * X1 X5 + 15 * X1

X5 X6 + 2 * X7. X8 + 1 * X7. X9. + 15 * X10 + 30 * X11 + 60 * X11 +
30 * X7 X3X13 + 3 * X7 X3 + 2 * X7 X14 + 10 * X7 X5
with: X1 = farmer routing (1) versus non-farmer routing (0)
X2 = first delivery: yes (1) versus no (0)
X3 = returned goods: yes (1) versus no (0)
X4 = returned reusable containers: yes (1) versus no (0)
X5 = cash payment: yes (1) or not (0)
X6 = customer is farmer or grower: yes (1) or no (0)
X7 = non-farmer (1) versus farmer (0)
X8 = number of pallets
X9 = number of packs
X10 = garden center: yes (1) or no (0)
X11 = large distribution center: yes (1) or no (0)
X12 = hypermarket Makro or Leroy Merlin: yes (1) or no (0)
X13 = appointment is made, so return code is known: no (1) or yes (0)

6
In programming language of Acorn this equation was written as: IF([CustomerType] = "Farmer", 5 +
IF([FirstOrder] = "Yes", 5, 0) + IF([Returns] <> 0, 5, 0) + IF([ReusableContainers] <> 0, 10, 0), 2 * [Pallets] + 1
* [Packs] + LOOKUP([CustomerType], "DIY", IF([Customer] = "Makro" OR [Customer] = "Leroy Merlin", 60,
30), "Garden Centre", 15, 0) + IF([Returns] <> 0, IF([Appointment] = "Yes", 3, 30), 0) + 2 * [EmptyPallets]) +
IF([CashPayment] = "Yes", IF([Customer Type] = "Farmer" OR [CustomerType] = "Grower", 20, 5), 0)
 29

X14 = empty pallets to return: yes (1) versus no (0)

The time equation fully captures the complicated structure of this activity. Many time drivers

affect the time for this activity. Some of them (X10, X11 and X12) only have a main effect on

the activity time (garden centers, large distribution centers, Makro or Leroy Merlin requiring

large waiting times). Most drivers have an interaction effect on the activity time. This means

that the time effect of the driver depends on the value of one or more other drivers. For

example: The time of a drop is not always 5 minutes per order in the farmers routing. When

this is a first delivery, an additional 5 minutes time is needed (two way interaction X1.X2

between “farmer versus non-farmers routing” (X1) and “first delivery” (X2). When no

appointment was made for the non-farmers, who have goods to return, the additional time can

be modeled by a three way interaction X7.X3.X13 between the three indicator variables ”non-

farmer versus farmer” (X7), “goods to return versus no goods to return” (X3) and “no return

code versus return code” (X13).

Also for this activity, sensitivity analysis shows that the activity time per drop significantly

differs depending on the characteristics of the delivery. For example an order shipped to a

farmer, no return of goods or reusable containers, no cash payment requires 5 minutes time

per drop. An order shipped to a farmer, normal delivery, goods and reusable containers to be

returned, cash payment requires 40 minutes time per drop. An order shipped on pallets to a

garden center, 10 pallets, goods to be returned but appointment has been made, no pallets to

return, no cash payment requires 38 minutes per drop. An order shipped on pallets to a

garden center, 10 pallets, goods to be returned, no appointment has been made, no goods to

return, cash payment requires 78 minutes per drop.

 30

6.5 Analysis of Model Complexity at Sanac

Studying the activities as described above (among others) in the time-driven ABC system of

Sanac reveals that logistics activities show different types of complexities.

1. Activities are not homogeneous and contain different subtasks, each having their own

different time driver. For instance, different drivers influence the time for unloading,

waiting, loading goods to return.

2. Necessity of different subtasks depends on the characteristics of the order, the

customer, etc. For instance, the commercial talk to the farmer (subtask) is only done

for a first time delivery in the farming routing (characteristic of the order) .

3. Time drivers of subtasks depend on the characteristics of the customers, the orders,

etc. For instance, unloading for farmers is depending on the number of drops, whereas

for non-farmer it depends on the number of pallets, or packs.

4. The required time per unit time driver depend on the characteristics of the customer,

the order, etc. For instance, time for cash payment is depending on the type of

customer (garden center, farmer, grower).

5. Activities (or subtasks within activities) are interdependent. For instance, calling for a

return code is needed when no appointment was made.

These complexities are translated in the time equations by including multiple time drivers and

interactions of time drivers, what will be discussed in paragraph 6.7. But first, we discuss

why the traditional ABC system could not capture these kinds of complexities.
 31

6.6 Why was traditional ABC not able to capture the identified complexities?

There are two ways to deal with the identified complexities in a traditional ABC model: (1)

simply ignoring the multiple time drivers per activity; or (2) creating a separate activity for

each subtask. We describe why none of these solutions was working out at Sanac. When

ignoring the multiple time drivers, the traditional ABC model worked with an activity driver

rate for each of the 106 identified activities, without including time drivers. For instance, for

the order entry activity, in traditional ABC method a cost per order entry is calculated without

considering the specific characteristics of that order. The same applies for the delivery cost

per drop, the reception cost per incoming delivery, the planning cost per trip, the

administration cost per product, the meeting cost per supplier, etc. This solution is still

keeping in mind the hierarchy of activities (unit, batch, product, customer, supplier), which is

much better than a traditional cost system with only volume drivers (such as turn over).

However, considering only an “average” activity driver rate (without time equation) resulted

for Sanac in distorted product cost information, because of the high diversity in orders,

customers, suppliers, routings, drop-offs and order lines. For instance, taking an average cost

per drop-off in traditional ABC is “undercosting” the deliveries that are consuming a lot of

resources (because of long waiting time, cash payments, many goods to return) and is

“overcosting” the deliveries that are not consuming a lot of resources (because of no waiting

time, no cash payments, no goods to return). As will be explained further (see Table 4), this

traditional ABC method created distortions for about 54% of the total activity costs by not

including multiple time-drivers.

The company then tried for the second solution, i.e. splitting-up the 106 logistics activities

into smaller activities representing the different subtasks. This resulted in 330 activities

(subtasks), as shown in Table 3. This large number of activities in the activity database made
 32

resource assignment to the activities extremely difficult. Taking the example of receiving

incoming goods, under the traditional ABC method Sanac needed to calculate the unloading

cost per pallet received, the checking cost per order line on the pallet received, the loading

cost per pallet to return, the loading cost per pallet to return with reusable containers, the

storing cost for one homogeneous pallet, the storing cost for one heterogeneous pallet. For

each of these activities the normal volume of the activity drivers needed to be estimated,

which was impossible. As you can see in Table 3, the proliferation of activities under the

traditional ABC model is mainly caused by the order, route and drop-off related activities.

The high increase in supplier-related activities is caused by a lot of subtasks that model the

negotiation style of the many suppliers.

Insert Table 3 about here.

6.7 Incorporation of the Complexities in the Time-Driven ABC model at Sanac

In order to explore whether the time-driven ABC model was able to capture the different kind

of complexities at Sanac, we wrote each of the 106 time equations in mathematical

representation as explained in paragraph 4. A summary is provided in Table 4. You see that

one time driver drives about 45% of the total resources. Around 54% of all resources needed

more than one time driver to explain the reality of the logistic operations. Table 5 shows that

about 30% of the activities contain 2 different time drivers and 31% contain more than 3 time

drivers. These multiple time drivers were needed because the activities were not

homogeneous.

Insert Table 4 about here.

 33

Insert Table 5 about here.

Table 6 provides an overview of the number of subtasks that were included in the time-driven

ABC model at Sanac. The strength of time-driven ABC is to include more than one variable

(constant or time driver) in the time equations. Two, three and four terms seem to be most

commonly used at Sanac, consuming respectively 15%, 9% and 15% of the resources. In

terms of the number of activities 31% contain two terms, whereas 4% contain 3 terms and

11% contain 4 terms. Remarkable is that 9% of the time equations contain 10 or more terms.

By including more than two terms, Sanac was able to cope with complex operations,

containing different subtasks with different unit time requirements.

Insert Table 6 about here.

Subtasks with different type of time drivers, depending on the characteristics of the event, ask

for two-way or three-way interaction terms. As shown in Table 7, 23% of the equations

contained a two-way interaction effect and 8% contained a three-way interaction effect. This

represented together 21% of the resource costs.

Insert Table 7 about here.

In sum, the time-driven ABC model at Sanac contained 106 time equations, 31% of the time

equations contained interaction terms (representing 21% of total costs), 61% of the time

equations contained multiple drivers (representing 54% of total costs). About 74% of the
 34

time equations included more than one term in the equation. These type of complexities

could easily be modeled in the time-driven ABC model at Sanac.

7 Conclusions

The case study shows that a time-driven ABC system was needed for this company. Similar

to the literature, this company found it too complex to include the high diversity of order-

related activities in a traditional ABC model and too time consuming for maintaining and

running the model on a monthly basis. Specific to this company was the seasonal trend in the

sales, asking for capacity utilization reports for the different departments. Additionally, this

company faced a strong diversity in resource consumptions by the various suppliers,

customers, orders, drops and routes that could not be captured in a traditional ABC model.

The examples of time equations for the reception of incoming goods and the delivery of

goods to customers showed that different kind of complexities were found in the logistics

operation. (1) Activities are not homogeneous and contain different subtasks, each having

their own different time driver. (2) The necessity of different subtasks depends on the

characteristics of the order, the type of customer, etc. (3) Time drivers of subtasks depend on

the characteristics of the order, the type of customer, etc. (4) The required time per unit time

driver is a function of the characteristics of the order, the type of customer, etc. (5) Subtasks

of activities are depending on the outcome of other activities (interdependency).

When using traditional ABC it was difficult to take these types of complexities into account.

The problem of different subtasks with different time drivers, situation dependent tasks,

drivers and costs per unit time driver can be solved by activity splitting. Also

interdependencies could be solved this way, but this method leads to a large number of

activities in the activity database, which makes resource assignment to the activities
 35

extremely difficult, less accurate and almost impossible to update. For Sanac this would lead

to 330 different activities, compared to 106 in the time-driven ABC approach.

Time-Driven ABC was able to capture the different types of complexities at Sanac. When

analyzing the time equations, we found that 64% of all activities contained more than one

term in the time equation, consuming 54% of all costs. Almost one third of the activities

asked for interaction terms, representing 21% of all resources. These quantifications show

that the time-driven ABC approach was needed at Sanac to provide accurate cost information.

Because of the high competitive pressure they could not tolerate to have 50% of their

resources misallocated to the different cost objects.

Many research opportunities are available for future studies.

 36

8 References

Anderson S., Hesford J., Young M., 2002, Factors influencing the Performance of Activity-
Based Costing Teams: A field study of ABC Model Development Time in the Automobile
Industry, Accounting, Organizations and Society, Volume 27, p. 195-211

Armstrong P., 2002, The Costs of Activity-Based Management, Accounting, Organizations

and Society, Volume 27, p. 99-120

Chen I.J., Paulraj A., 2004, Understanding Supply Chain Management: Critical Research and
a Theoretical Framework, International Journal of Production Research, January, 42 (1),
131-163

Cobb I., Innes J., Mitchell F., 1994, Activity-Based Costing Problems: The British
Experience, Advances in Management Accounting, Volume 2, p. 63-83

Cokins G., 1999, Are All of Your Trading Partners “Worth It” to You?, White Paper,
http://cokins.ascet.com

Cooper R, 1988a, The rise of Activity-based Costing: Part one: What is an Activity-based
Cost System?, Journal of Cost Management, Summer 1988, p. 45-54

Cooper R., 1988b, The rise of Activity-based Costing: Part two: When do I need an Activity-
based Cost System?, Journal of Cost Management, Fall 1988, p. 41-48

Cooper R., 1989a, The rise of Activity-based Costing: Part three: How many Cost Drivers Do
You need? And How Should you select them?, Journal of Cost Management, Winter 1989, p.
34-46

Cooper R., 1989b, The rise of Activity-based Costing: Part four: What do Activity-based Cost
System look like?, Journal of Cost Management, Spring 1989, p. 34-46

Cooper R., 1990, Implementing an Activity-based Cost System, Journal of Cost Management,
Spring 1990, p. 33-42

Cooper R., Kaplan R., 1988, Measure Costs Right: Make the right Decisions, Harvard
Business Review, May 1988, p. 96-103

Cooper R., R.S. Kaplan, L.S. Maisel, E. Morrissey, R.M. Oehm, 1992, Implementing activity-
based cost management: Moving from Analysis to Action, Montvale NJ, Institute of
Management Accountants

Datar S., Gupta M., Aggregation, Specification and Measurement Errors in Product Costing,
The Accounting Review, Volume 69, N° 4, October 1994, p. 567-591

Foster G., M. Gupta, 1990, Manufacturing Overhead Cost Driver Analysis, Journal of
Accounting and Economics, Volume 12, N° 1/2/3, 309-337

Foster G., Gupta M., Sjoblom L., 1996, Customer Profitability Analysis: Challenges and New
Directions, Journal of Cost Management, Spring 1996, p. 5-17
 37

Foster G., Swenson D., 1997, Measuring the Success of Activity-Based Cost Management
and Its Determinants, Journal of Management Accounting Research, Volume 9, p. 109-141

Innes J., Mitchell F., Cobb I., 1992, Activity-based costing: Problems in Practice, London,
CIMA

Kaplan R., Anderson S., 2004, Harvard Business Review, Time-Driven Activity-Based
Costing, November 2004, p. 131-138

Kaplan R., Cooper R., 1998, Cost & Effect, Using Integrated Cost Systems to Drive
Profitability and Performance, Harvard Business School Press

Lin B., Collins J., Su R., 2001, Supply Chain Costing: an Activity-based Perspective,
International Journal of Physical Distribution and Logistics Management, 2001, Volume 31,
N° 9/10, p. 702-713

Norek C., Pohlen T., 2001, Cost Knowledge: A foundation for Improving Supply Chain
Relationships, International Journal of Logistics Management, 2001, Volume 12, N° 1, p. 37-
51

Otley D. T., Berry A.J., 1994, Case study research in Management Accounting and Control,
Management Accounting Research, 1994, volume 5, p. 45-65

Pohlen T., La Londe B., 1994, Implementing Activity-Based Costing (ABC) in Logistics,
Journal of Business Logistics, Volume 15, N° 2, p. 1-23

Ryan, B., Scapens R., Theobold M., 2003, Research Method and Methodology in Finance
and Accounting, Thomson, London

Scapens B., 1990, Researching Management Accounting Practice: The role of Case Study
Methods, British Accounting Review, 1990, Volume 22, p. 259-281

Themido I, Arantes A., Fernandes C., Guedes AP, 2000, Logistic Costs Case Study: an ABC
Approach, Journal of the Operational Research Society, 2000, Volume 51, p. 1148-1157
 38

Figure 1: Traditional Activity-Based Costing

Activity-Based Cost Systems trace resource expenses to activities and use

activity cost drivers for tracing activity costs to objects.

Resource Pool Administration

Resources expenses Salary Depreciation ICT ...

Resource driver

Ordering Sales
Activities Order entry ... ...
material invoicing

No. of No. of No. of sales ...

Activity Cost Driver orders purchases

Activity Cost Driver Rate =

total cost of the activity
total volume activity cost driver $/order $/purchase $/invoice ...

Direct material
Product A Product B Product C
Direct labor

Source: Based on Kaplan & Cooper (1998)

 39

Figure 2: Time-Driven Activity-Based Costing

Time-Driven Activity-Based Cost Systems trace costs of resource pools to

objects, based on the outcomes of the time equations per activity

Resources expenses Salary Depreciation ICT ...

Resource driver

Resource pools Administration Warehouse ...

Unit cost per resource pool =

total cost of resource pool C1= $/min C2= $/min Cn= $/min
practical capacity

Order entry Ordering Sales

Activities material invoicing

tactivity1, k = β 0 + β1.Χ1 + β 2 .Χ 2 + ... + β p .Χ p

... ...
Time equations with tactivity 2, k = β 0 + β1.Χ1 + β 2 .Χ 2 + ... + β p .Χ p
time drivers tactivity 3, k = β 0 + β1.Χ1 + β 2 .Χ 2 + ... + β p .Χ p

Time spent multiplied by unit ∑ ∑ tj,k * C1 ∑ ∑ tj,k * C2 ∑ ∑ tj,k * Cn

cost per resource

Direct material
Product A Product B Product C
Direct labor
 40

Figure 3: Profitability Analysis for the different Cost Objects

Companywide

BU

Customers

Orders Suppliers

Products
Order lines

Source: Internal Documents at Ranger

 41

Table 1: Time Drivers for the Activity “Reception of Incoming Goods”

Subtask Time Driver Time consumption per

unit time driver
Dock assignment Number of deliveries 1 minute per delivery
Input general information Number of deliveries 1 minute per delivery
Input order line information Number of purchase order lines 10 seconds per order
per delivery line
Unloading truck Number of pallets per delivery 2,5 minute per pallet
Communicating information Number of deliveries 1 minute per delivery
Returning reusable containers Number of pallets with containers 10 minutes per pallet
per delivery with containers
Returning pallets Number of staples of returned 3 minutes per staple
pallets per delivery pallets to return
Checking and storing Number of homogeneous pallets 1 minute per
per delivery homogeneous pallet
Checking and storing Number of heterogeneous pallets 15 minutes per
per delivery heterogeneous pallet
Signing off the voucher Number of deliveries 1 minute per delivery
 42

Table 2: Time Drivers for the Activity “Drop off goods to customers”

Subtask Time driver Time consumption per

unit time driver
Dropping the goods • Farmer or non-farmers’ routing:
o Number of drops, in case of • 5 minutes per drop
delivery to farmers for farmers
o Number of pallets, in case of • 2 minutes per pallet
delivery on pallets to non- for drop to non-
farmers farmers
o Number of collies, in case of • 1 minute per pack
delivery in collies to non- for drop to non-
farmers farmer
• First delivery or not, in case of • 10 minutes per drop
delivery to farmers to a new farmer.
Take back returned • Returned goods or not:
goods o Number of drops • 5 minutes per drop
for farmers
• 3 minutes per drop
for a non-farmer

o Appointment being made or not, • 30 minutes per drop

in case of returns from non- to non-farmer if no
farmers appointment was
made
Take back reusable • Reusable containers to take back or
containers not:
o Number of drops, in case of • 10 minutes per drop
returns from farmers to farmer

Waiting • For non-farmers: type of customer:

o Garden centre or not • 15 minutes per drop
to garden centre
o Large distributor or not • 30 minutes per drop
to large distributor
o Leroy Merlin, Makro or not • 60 minutes per drop
to Makro or Leroy
Merlin
Take back pallets • Pallets to take back or not:
o Only for non-farmers: the • 2 minutes per drop
number of drops to non-farmer
Cash payment • Cash payment or not:
o For farmer’s routing, type of
customer:
• Farmer • 15 minutes per drop
• Grower • 15 minutes per drop
• Other • 5 minutes per drop
o For non-farmer’s routing: • 10 minutes per drop
 43

Table 3: Number of Activities in the Time-Driven ABC Model at Sanac

Number of Activities in Number of % Increase in

Time
Resources in activities in % of total for activities as-if activities
equations for
% of total Time-Driven Time-Driven traditional
each …
ABC (1) ABC ABC (2) (2) – (1)/(2)
Supplier 10% 28 26% 133 + 375%
Product 4% 14 13% 20 + 43%
Customer 9% 29 27% 45 + 55%
Order 48% 23 22% 87 + 278%
Route 12% 4 4% 13 +225%
Drop-off 10% 5 5% 23 +360%
Order line 7% 3 3% 9 + 200%
100% 106 100% 330 + 211%
 44

Table 4: Summary of the Time Equations found at Sanac

Time Equation Specifications Resources # of Activities

in % of activities in % of
total total
t j, k = β 0 1% 5 5%

t j , k = β1.Χ1 45% 33 31%

t j , k = β 0 + β1.Χ1 0% 3 3%

t j , k = β1.Χ1 + β 2 .Χ 2 10% 15 14%

t j , k = β1.Χ1 + β 2 .Χ1Χ 2 5% 5 14%

t j , k = β 0 + β1.Χ1 + β 2 .Χ 2 5% 2 2%

t j , k = β1.Χ1 + β 2 .Χ 2 + β 3 .Χ 3 4% 2 2%

t j , k = β 0 + β1.Χ1 + β 2 .Χ 2 + β 3 .Χ 3 8% 4 4%

t j , k = β1.Χ1 + β 2 .Χ1Χ 2 + β 3 .Χ1Χ 3 + β 4 .Χ1Χ 4 1% 3 3%

t j , k = β1.Χ1Χ 2 + β 2 .Χ1Χ 2 Χ 3 + β 3 .Χ1Χ 2 Χ 4 + 4% 3 3%

β 4 .Χ1Χ 2 Χ 5
t j , k = β1.Χ1Χ 2 + β 2 .Χ1Χ 2 Χ 3 + β 3 .Χ 4 + β 4 .Χ 5 2% 1 1%

t j , k = β1.Χ1 + β 2 .Χ 2 + β 3 .Χ 3 + β 4 .Χ 4 + β 5 .Χ 5 1% 2 2%

Time equations with more than 5 time drivers and 15% 18 17%
more than 5 terms, on average 3,2 two-way
interactions and 0,9 three-way interaction per equation
100% 106 100%
 45

Table 5: Summary of the Number of Time Drivers in the Time Equations at Sanac

Resources in # of Activities in
Time equations with
% of total Activities % of total
Only ß0 1% 5 5%
1 time driver (with or without ß0) 46% 36 34%
2 time drivers (with or without ß0) 20% 32 30%
3 time drivers (with or without ß0) 12% 6 6%
4 time drivers (with or without ß0) 1% 3 3%
5 time drivers (with or without ß0) 6% 6 6%
7 time drivers (with or without ß0) 6% 8 7%
More than 7 time drivers (with or without ß0) 8% 10 9%
100% 106 100%
 46

Table 6: Summary of the Number of Terms in the Time Equations at Sanac

Resources in # of Activities in
Time equations with
% of total Activities % of total
1 term in the time equation 46% 38 36%
2 terms in the time equation 15% 33 31%
3 terms in the time equation 9% 4 4%
4 terms in the time equation 15% 11 11%
5 terms in the time equation 1% 4 4%
7 terms in the time equation 3% 3 3%
9 terms in the time equation 3% 3 3%
10 terms or more in the time equation 8% 10 9%
100% 106 100%
 47

Table 7: Summary of the Type of Interactions in the Time Equations at Sanac

Resources in # of Activities in
Time equations with
% of total Activities % of total
No interactions 79% 73 69%
Two-way interactions: 1 or more 9% 25 23%
Three-way interactions: 1 or more 12% 8 8%
Higher order interactions 0% 0 0%
100% 106 100%
 FACULTEIT ECONOMIE EN BEDRIJFSKUNDE
HOVENIERSBERG 24
9000 GENT Tel. : 32 - (0)9 – 264.34.61
Fax. : 32 - (0)9 – 264.35.92

WORKING PAPER SERIES 14

04/259 D. VANDAELE, P. GEMMEL, Development of a measurement scale for business-to-business service quality:
assessment in the facility services sector, September 2004, 30 p.

04/260 F. HEYLEN, L. POZZI, J. VANDEWEGE, Inflation crises, human capital formation and growth, September 2004, 23
p.

04/261 F. DE GRAEVE, O. DE JONGHE, R. VANDER VENNET, Competition, transmission and bank pricing policies:
Evidence from Belgian loan and deposit markets, September 2004, 59 p.

04/262 B. VINDEVOGEL, D. VAN DEN POEL, G. WETS, Why promotion strategies based on market basket analysis do
not work, October 2004, 19 p. (forthcoming in Expert Systems with Applications, 2005)

04/263 G. EVERAERT, L. POZZI, Bootstrap based bias correction for homogeneous dynamic panels, October 2004, 35 p.

04/264 R. VANDER VENNET, O. DE JONGHE, L. BAELE, Bank risks and the business cycle, October 2004, 29 p.

04/265 M. VANHOUCKE, Work continuity constraints in project scheduling, October 2004, 26 p.

04/266 N. VAN DE SIJPE, G. RAYP, Measuring and Explaining Government Inefficiency in Developing Countries, October
2004, 33 p.

04/267 I. VERMEIR, P. VAN KENHOVE, The Influence of the Need for Closure and Perceived Time Pressure on Search
Effort for Price and Promotional Information in a Grocery Shopping Context, October 2004, 36 p. (published in
Psychology & Marketing, 2005).

04/268 I. VERMEIR, W. VERBEKE, Sustainable food consumption: Exploring the consumer attitude – behaviour gap,
October 2004, 24 p.

04/269 I. VERMEIR, M. GEUENS, Need for Closure and Leisure of Youngsters, October 2004, 17 p.

04/270 I. VERMEIR, M. GEUENS, Need for Closure, Gender and Social Self-Esteem of youngsters, October 2004, 16 p.

04/271 M. VANHOUCKE, K. VAN OSSELAER, Work Continuity in a Real-life Schedule: The Westerschelde Tunnel,
October 2004, 12 p.
04/272 M. VANHOUCKE, J. COELHO, L. V. TAVARES, D. DEBELS, On the morphological structure of a network,
October 2004, 30 p.

04/273 G. SARENS, I. DE BEELDE, Contemporary internal auditing practices: (new) roles and influencing variables.
Evidence from extended case studies, October 2004, 33 p.

04/274 G. MALENGIER, L. POZZI, Examining Ricardian Equivalence by estimating and bootstrapping a nonlinear dynamic
panel model, November 2004, 30 p.

04/275 T. DHONT, F. HEYLEN, Fiscal policy, employment and growth: Why is continental Europe lagging behind?,
November 2004, 24 p.
04/276 B. VINDEVOGEL, D. VAN DEN POEL, G. WETS, Dynamic cross-sales effects of price promotions: Empirical
generalizations, November 2004, 21 p.

04/277 J. CHRISTIAENS, P. WINDELS, S. VANSLEMBROUCK, Accounting and Management Reform in Local

Authorities: A Tool for Evaluating Empirically the Outcomes, November 2004, 22 p.

04/278 H.J. SAPIENZA, D. DE CLERCQ, W.R. SANDBERG, Antecedents of international and domestic learning effort,
November 2004, 39 p.

04/279 D. DE CLERCQ, D.P. DIMO, Explaining venture capital firms’ syndication behavior: A longitudinal study, November
2004, 24 p.
 FACULTEIT ECONOMIE EN BEDRIJFSKUNDE
HOVENIERSBERG 24
9000 GENT Tel. : 32 - (0)9 – 264.34.61
Fax. : 32 - (0)9 – 264.35.92

WORKING PAPER SERIES 15

04/280 T. FASEUR, M. GEUENS, Different Positive Feelings Leading to Different Ad Evaluations: The Case of Cosiness,
Excitement and Romance, November 2004, 17 p.

04/281 B. BAESENS, T. VAN GESTEL, M. STEPANOVA, D. VAN DEN POEL, Neural Network Survival Analysis for
Personal Loan Data, November 2004, 23 p. (Forthcoming in Journal of the Operational Research Society)

04/282 B. LARIVIÈRE, D. VAN DEN POEL, Predicting Customer Retention and Profitability by Using Random Forests and
Regression Forests Techniques, December 2004, 30 p. (Forthcoming in Expert Systems with Applications, 2005).

05/283 R. I. LUTTENS, E. OOGHE, Is it fair to “make work pay”?, January 2005, 28 p.

05/284 N. A. DENTCHEV, Integrating Corporate Social Responsibility In Business Models, January 2005, 29 p.

05/285 K. FARRANT, G. PEERSMAN, Is the exchange rate a shock absorber or a source of shocks? New empirical
evidence, January 2005, 26 p. (forthcoming Journal of Money Credit and Banking, 2005)

05/286 G. PEERSMAN, The relative importance of symmetric and asymmetric shocks and the determination of the
exchange rate, January 2005, 24 p.

05/287 C. BEUSELINCK, M. DELOOF, S. MANIGART, Private Equity Investments and Disclosure Policy, January 2005, 44
p.

05/288 G. PEERSMAN, R. STRAUB, Technology Shocks and Robust Sign Restrictions in a Euro Area SVAR, January
2005, 29 p.

05/289 H.T.J. SMIT, W.A. VAN DEN BERG, W. DE MAESENEIRE, Acquisitions as a real options bidding game, January
2005, 52 p.

05/290 H.T.J. SMIT, W. DE MAESENEIRE, The role of investor capabilities in public-to-private transactions, January 2005,
41 p.

05/291 A. VEREECKE, S. MUYLLE, Performance Improvement Through Supply Chain Collaboration: Conventional Wisdom
Versus Empirical Findings, February 2005, 33 p.

05/292 A. PRINZIE, D. VAN DEN POEL, Incorporating sequential information into traditional classification models by using
an element/position- sensitive SAM, February 2005, 32 p. (Forthcoming in Decision Support Systems).

05/293 D. DEBELS, M. VANHOUCKE, A Decomposition-Based Heuristic For The Resource-Constrained Project Scheduling
Problem, February 2005, 25 p.

05/294 D. DEBELS, M. VANHOUCKE, A Bi-Population Based Genetic Algorithm for the Resource-Constrained Project
Scheduling Problem, February 2005, 10 p.

05/295 W. VANDEKERCKHOVE, N.A. DENTCHEV Network perspective on stakeholder management: Facilitating entre-
preneurs in the discovery of opportunities, February 2005, 31 p.

05/296 N. MORAY, B. CLARYSSE, Institutional Origin and Resource Endowments to Science-Based Entrepreneurial Firms:
A European Exploration, February 2005, 39 p.

05/297 B. CLARYSSE, M. KNOCKAERT, A. LOCKETT, How do Early Stage High Technology Investors Select Their
Investments?, March 2005, 38 p.

05/298 A. PRINZIE, D. VAN DEN POEL, Constrained optimization of data-mining problems to improve model performance:
A direct-marketing application, March 2005, 32 p. (forthcoming in Expert Systems with Applications, 2005).

05/299 B. LARIVIÈRE, D. VAN DEN POEL, Investigating the post-complaint period by means of survival analysis, March
2005, 25 p.
 FACULTEIT ECONOMIE EN BEDRIJFSKUNDE
HOVENIERSBERG 24
9000 GENT Tel. : 32 - (0)9 – 264.34.61
Fax. : 32 - (0)9 – 264.35.92

WORKING PAPER SERIES 16

05/300 D. VANDAELE, P. GEMMEL, Impact of Support Services on End-users’ Satisfaction, March 2005, 28 p.

05/301 T. DEMUYNCK, L. LAUWERS, Nash rationalizability of collective choice over lotteries, April 2005, 16 p.

05/302 R. BRACKE, M. DE CLERCQ, Implementing Extended Producer Responsibility in Flemish Waste Policy:
Evaluation of the Introduction of the Duty of Acceptance, April 2005, 24 p.

05/303 Ph. DE MOOR, I. DE BEELDE, Environmental Auditing and the Role of the Accountancy Profession: A Literature
Review, April 2005, 34 p.

05/304 G. SARENS, I. DE BEELDE, Internal Auditor’s Perception about their Role in Risk Management Comparison
between Belgian and US Companies, April 2005, 25 p.

05/305 Ph. VAN CAUWENBERGHE, I. DE BEELDE, On the IASB Comprehensive Income Project, Neutrality of Display
and the Case for Two EPS Numbers, May 2005, 29 p.

05/306 P. EVERAERT, G. SARENS, Outsourcing bij Vlaamse Ondernemingen: een Exploratief Onderzoek, May 2005,
36 p.

05/307 S. CLAEYS, G. LANINE, K. SCHOORS, Bank Supervision Russian Style: Rules vs Enforcement and Tacit
Objectives, May 2005, 60 p.

05/308 A. SCHOLLAERT, D. VAN DE GAER, Boycotts, power politics or trust building: how to prevent conflict?, June
2005, 34 p.

05/309 B. MERLEVEDE, K. SCHOORS, How to Catch Foreign Fish? FDI and Privatisation in EU Accession Countries,
June 2005, 32 p.

05/310 A. HEIRMAN, B. CLARYSSE, The Imprinting Effect of Initial Resources and Market Strategy on the Early Growth
Path of Start-Ups, June 2005, 38 p.

05/311 A. MAES, G. POELS, J. GAILLY, R. PAEMELEIRE, Measuring User Beliefs and Attitudes towards Conceptual
Models: A Factor and Structural Equation Model, June 2005, 25 p.

05/312 S. VANDEVOORDE, M. VANHOUCKE, A Comparison of Different Project Duration Forecasting Methods using
Earned Value Metrics, June 2005, 18 p.

05/313 K. BAEYENS, T. VANACKER, S. MANIGART, Venture capitalists’ selection process: the case of biotechnololy
proposals, June 2005, 23 p.

05/314 M. VANHOUCKE, D. DEBELS, The discrete time/cost trade-off problem under various assumptions. Exact and
heuristic procedures, June 2005, 32 p.

05/315 G. POELS, A. MAES, F. GAILLY, R. PAEMELEIRE, User Attitudes towards Pattern-Based Enterprise
Information Models: A Replicated Experiment with REA Diagrams, July 2005, 32 p.

05/316 B. MAENHOUT, M. VANHOUCKE, An Electromagnetic Meta-Heuristic for the Nurse Scheduling Problem, July
2005, 34 p.

05/317 M. VANHOUCKE, S. VANDEVOORDE, A simulation and evaluation of earned value metrics to forecast the
project duration, July 2005, 30 p.

05/318 I. VERMEIR, M. GEUENS, Need for Closure and Youngsters’ Leisure Time Preferences, July 2005, 25 p.
 FACULTEIT ECONOMIE EN BEDRIJFSKUNDE
HOVENIERSBERG 24
9000 GENT Tel. : 32 - (0)9 – 264.34.61
Fax. : 32 - (0)9 – 264.35.92

WORKING PAPER SERIES 17

05/319 D. VANTOMME, M. GEUENS, S. DEWITTE, How to Portray Men and Women in Advertisements? Explicit and
Implicit Evaluations of Ads Depicting Different Gender Roles, July 2005, 34 p.

05/320 B. WEIJTERS, M. GEUENS, Evaluation of age-related labels by senior citizens, July 2005, 27 p.

05/321 S. STEENHAUT, P. VANKENHOVE, An Empirical Investigation of the Relationships among a Consumer’s

Personal Values, Ethical Ideology and Ethical Beliefs, July 2005, 36 p.

05/322 V. VANSTEEGER, The current state of accounting harmonization: impediments to and benefits from
harmonization, August 2005, 30 p.

05/323 G. POELS, F. GAILLY, A. MAES, R. PAEMELEIRE, Object Class or Association Class? Testing the User Effect
on Cardinality Interpretation, August 2005, 14 p.

05/324 G. BUCKINX, G. VERSTRAETEN, D. VAN DEN POEL, Predicting Customer Loyalty Using The Internal
Transactional Database, August 2005, 33 p.

05/325 M. KNOCKAERT, A. LOCKETT, B. CLARYSSE, M. WRIGHT, Do human capital and fund characteristics drive
follow-up behaviour of early stage high tech VCs?, August 2005, 27 p.

05/326 H. DEWITTE, E. VERHOFSTADT, E. OMEY, Testing Karasek’s learning- and strain hypothesis on young workers
in their first job, August 2005, 27 p.

05/327 A. KARAS, K. SCHOORS, Heracles or Sisyphus? Finding, cleaning and reconstructing a database of Russian
banks, September 2005, 26 p.

05/328 M. BUELENS, D. BOUCKENOOGHE, D. DE CLERCQ, A. WILLEM, An Assessment of Validity in Small

Business and Entrepreneurship Research, September 2005, 33 p.

05/329 G. LANINE, R. VANDER VENNET, Failure prediction in the Russian bank sector with logit and trait recognition
models, August 2005, 31 p.

05/330 W. BUCKINX, D. VAN DEN POEL, Assessing and exploiting the profit function by modeling the net impact of
targeted marketing, September 2005, 30 p.

05/331 J. CHRISTIAENS, P. WINDELS, Management Reform in Flemish Local Authorities: Testing the Institutional
Framework, September 2005, 24 p.