LIBRARY
Michigan State
University This is to certify that the

 

 

dissertation entitled

 

ANALYSIS OF PACKAGE LABELING:
AN EVALUATIVE TOOL TO COMPARE COST
STRUCTURES AND IMPLEMENTATION OF
POSTPONEMENT STRATEGIES

presented by

IREM AYDINSOY KIYAK

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Packaging

 

 

 

flMt/F/QWL

Major F’rofesso‘s‘SZJnature

October 27, 2003

 

Date

MSU is an Affirmative Action/Equal Opportunity Institution

.-n-.-g-¢-.-9-.-.-.-.-.-.-.-o-.-

C-.-l-o-I-C-.-u-I-.-I-O-I-.-O-O-.-o-o-o-O-n-o-O-I-0-o-0-0-0-I-l-I-O----. -

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DateDue.p65-p. 15

 

. - _ .lrI—J‘

ANALYSIS OF PACKAGE LABELING:
AN EVALUATIVE TOOL TO COMPARE COST STRUCTURES AND
IMPLEMENTATION OF POSTPONEMENT STRATEGIES

BY

Irem Aydinsoy Kiyak

A DISSERTATION
Submitted to
Michigan State University
In partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY

School of Packaging

2003

ABSTRACT
ANALYSIS OF PACKAGE LABELING:
AN EVALUATIVE TOOL TO COMPARE COST STRUCTURES AND

IMPLEMENTATION OF POSTPONEMENT STRATEGIES

BY

Irem Aydinsoy Kiyak

The package labeling function within the pharmaceutical industry
is examined for the purpose of developing a total cost simulation
model for evaluating postponement strategies. A combination
simulation-regression analysis is conducted using data obtained via the
case study interview method. Based on these findings, variables were
developed and tested in the model to determine the economic viability
of postponing labeling to the distribution center level.

Different types of economic structure drive two separate printing
technologies, namely plate (conventional) and plateless (digital).
Depending on the technology utilized, pharmaceutical companies may
generate cost savings through postponing the labeling function to the
distribution center level in the supply chain. Digital printing is
identified as an enabler to postponement clue to significantly reduced

setup times.

In conclusion, the study identifies and defines the product
factors that have an impact on the total cost of printing labels for the
pharmaceutical industry. Results indicate that the number of annual
production runs is the factor that has the largest impact on total cost
savings realized through the use of digital printing. Consequently, total
cost savings achieved through digital printing technologies become
increasingly greater as the number of annual production runs increase.
The second largest impact is the number of brands, with the number
of global markets and the frequency of marketing changes tied for the
next order of importance. Managerial implications are discussed where
labeling postponement becomes economically viable for a given
product line in the pharmaceutical industry, based on the defined

product factors.

Copyright by
IREM AYDINSOY KIYAK

2003

This dissertation is dedicated to my mother Saadet A ydinsoy,
my father Halil Ibrahim A ydinsoy, and my husband T unga
Kiyak, for not letting me give up in hardship and for their
endless patience, deep love, unconditional support, and

confidence.

ACKNOWLEDGEMENTS

Writing a dissertation is a very demanding task. Continuous
understanding and support from one’s family and friends in addition to
considerable input from committee members, research participants,
and colleagues are critical components in completing a dissertation.

First and foremost, I would like to thank my committee members
at Michigan State University (MSU) starting with Dr. Robb Clarke,
Chairperson of the committee, Professor of Packaging, for his enduring
efforts, never-ending patience, and support. From the very first day,
Dr. Clarke took me under his wing and served as my guiding light
throughout the dissertation process. Dr. David Closs, Professor of
Marketing and Supply Chain Management, The John H. McConnell
Chair in Business Administration, was very instrumental in developing
the methodology and simulation. Dr. Closs’ expertise was much
needed and his contributions in the Packaging field are greatly
appreciated. I was very lucky to have Dr. Bruce Harte, Professor and
Director of the School of Packaging, in my committee. His dedication
to clarity in writing and investigative contributions helped greatly. I
would like to thank Dr. Tomas Hult, Professor of Marketing and Supply
Chain Management, who never tired of saying “you’ve GOT to finish”

during my many times of hardship. Dr. Hult’s much needed

vi

contributions in the statistical analysis are greatly appreciated. He was
my thunderous advocate and I owe him greatly. I would like to thank
Dr. Hugh Lockhart, Professor of Packaging, for his assistance in
assuring the quality of the dissertation. Thanks to Dr. Diana Twede,
Professor of Packaging, for keeping me in her radar screen at all times,
sending me articles, and her continuous supply of motivating words
“just sit down and do it,” I will not forget.

This research would not be possible had The John Henry
Company of Lansing, Michigan (JHC) not brought the business problem
to the School of Packaging. Dr. Shahriar Ghoddousi, The CEO and Vice
Chairman of JHC, deserves much credit for identifying the business
questions and seeking answers from the academia. Dr. Ghoddousi’s
academic background combined with his vast experience in the
industry was critical. Mr. Reha Emden, Director of Digital Packaging at
JHC dedicated precious long hours not only in the initial phases of the
study but throughout the whole process. Having completed a master's
thesis himself at the Rochester Institute of Technology, his
understanding of the theoretical concepts and immense knowledge in
the practice provided much insight. Mr. Emden’s commitment to the
packaging industry is greatly admired. Mr. Tim Schultz, Vice President
of JHC, provided access to resources that I could not have had

otherwise.

vii

Much business and academic insight were needed in designing
this study and the following people/organizations have helped me
obtain that and more: Dr. Laura Bix, Edward A. Church, Ray Dickinson,
Dr. Rosamarie Feliu, Alvin Mike Foster, Institute of Packaging
Professionals, Matthew Neumann, Perrigo Company, Alfred McKinlay,
Susan B. McNeely, Donald 0. Morris, Walt Sledzieski, Matthew
Thompson, Kurt Trombley, and Christine Urdiales.

Many thanks and debt are owed to Dr. S. Tamer Cavusgil,
University Distinguished Faculty, The John William Byington Endowed
Chair in Global Marketing, Professor of Marketing and International
Business, for giving me the opportunity to succeed as a higher
education administrator and always encouraging me to develop myself
both professionally and personally.

My employer, MSU’s International Business Center/ MSU-CIBER
(Center for International Business Education and Research) deserves
much appreciation and heartfelt thanks for partially funding this
research through supporting my educational assistance, and Dr.
Tomas Hult, Director of MSU-CIBER for giving me time off to work on
it. Thanks to Ms. Beverly Wilkins, Ms. Ronda Bunnell, Mr. Michael
Reed, and many of our graduate and student assistants who did not

have to, but DID put up with me, provided both emotional and

viii

administrative support, and assisted me in many aspects during my
stressful and needy times.

Thanks to Dr. Roger Calantone, The Eli Broad University
Professor, Marketing and Supply Chain Management at MSU, for his
past contributions in the simulation and modeling areas of marketing
and supply chain management. The simulation model used in this
study followed some of the foundations he had established.

Finally, without the deep love, understanding, and endless
support of my husband, Tunga Kiyak (soon to be Dr.), I could not have
accomplished this enduring task. In his own right, Tunga wrote yet
another dissertation in many more ways than helping me with the
statistical analysis, providing advice even in areas unfamiliar to him,
and cradling me through the puzzles and stresses of life. My parents,
Saadet and Ibrahim Aydinsoy, totally dedicated their lives to my (and
my little brother’s) education not just by providing their life savings to
my disposal for me to get the best education available but also
through their extensive tutorial and continuous emotional support.
Many nights I had as a pre-teen and a teenager were spent huddling
with my mother over a physics or a math homework with continuous
supply of fresh fruits peeled and sliced for me as I studied. Words are
just not enough, but I am most grateful and in debt for their lifelong

support and love.

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES
LIST OF EQUATIONS

CHAPTER I
INTRODUCTION
GENERAL OVERVIEW
BUSINESS PROBLEM
LIMITED PATENT PERIOD
REGULAR CONTENT CHANGES
SIGNIFICANT CUSTOMIZATION
BUSINESS PROBLEM SUMMARIZED
RESEARCH OBJECTIVES AND QUESTIONS
RESEARCH METHOD OVERVIEW
RESEARCH CONTRIBUTIONS

CHAPTER II

LITERATURE REVIEW
PRINTING TECHNOLOGIES
POSTPONEMENT-SPECULATION STRATEGIES
ACTIVITY- BASED COSTING
TOTAL COST APPROACH
ANALYSIS OF PACKAGE LABELING

CHAPTER III
MODEL DEVELOPMENT AND RESEARCH DESIGN
MODEL AND THEORY DEVELOPMENT
COST FACTORS
PRODUCT FACTORS
MODEL STRUCTURE
RESEARCH HYPOTHESES
METHODOLOGY AND RESEARCH DESIGN
CASE STUDY INTERVIEW
COST SIMULATION
CONCLUSION

xii

xiv

XV

AOLDmNOI-b-hNi—t

HH

16
20
31
34
36

4O
42
45
59
69
71
74
74
77
82

CHAPTER IV
DATA ANALYSIS
EFFECT OF FREQUENCY OF REGULATORY CHANGES
EFFECT OF FREQUENCY OF MARKETING CHANGES
EFFECT OF FREQUENCY OF PACKAGE ENGINEERING CHANGES
EFFECT OF VALUE OF PRODUCT-PACKAGE SYSTEM
EFFECT OF NUMBER OF BRANDS
EFFECT OF NUMBER OF GLOBAL MARKETS
EFFECT OF NUMBER OF ANNUAL PRODUCTION RUNS
MULTIPLE VARIABLE REGRESSION ANALYSIS
EXAMINATION OF RELATIVE IMPORTANCE OF PRODUCT FACTORS

CHAPTER V
RESULTS
EFFECT OF FREQUENCY OF REGULATORY CHANGES
EFFECT OF FREQUENCY OF MARKETING CHANGES
EFFECT OF FREQUENCY OF PACKAGE ENGINEERING CHANGES
EFFECT OF VALUE OF PRODUCT-PACKAGE SYSTEM
EFFECT OF NUMBER OF BRANDS
EFFECT OF NUMBER OF GLOBAL MARKETS
EFFECT OF NUMBER OF ANNUAL PRODUCTION RUNS
CONCLUSION

CHAPTER VI

DISCUSSION AND IMPLICATIONS
RESEARCH LIMITATIONS
FUTURE RESEARCH CONSIDERATIONS
MANAGERIAL IMPLICATIONS

APPENDIX A
INTERVIEW QUESTIONS

APPENDIX 8
FIRST 25 OBSERVATIONS IN SIMULATION MODEL

BIBLIOGRAPHY

xi

83
85
88
91
94
98
100
103
105
107

110
111
112
113
114
115
115
116
117

118
119
121
124

131

136

142

Table
Table
Table
Table
Table
Table
Table
Table
Table
Table
Table

Table

Table

Table

Table

Table
Table

Table
Table
Table

LIST OF TABLES

1.1 -Postponement-Speculation for Label Printing

2.1 —Reasons for Digital Printing in the Industry

2.2 -Types of Postponement

2.3 —Postponement-Speculation Matrix

2.4 —Postponement—Speculation Decision Factors

3.1 -Features Impacting Cost of Printing Labels

3.2 —Product Factors Affecting Cost of Printing Labels

3.3 —Traditional Cost Model of Label Printing

3.4 —Activity-Based Cost Model of Label Printing

3.5 —Summary of Total Cost Formulas and Variables

3.6a-Product Factors, Associated Ranges, and
Their Relation to Cost Factors

3.6b- Product Factors, Associated Ranges, and
Their Relation to Cost Factors

3.6c—Product Factors, Associated Ranges, and
Their Relation to Cost Factors

3.6d- Product Factors, Associated Ranges, and
Their Relation to Cost Factors

3.6e- Product Factors, Associated Ranges, and
Their Relation to Cost Factors

3.7 -Effect of Product Factors on Cost Factors

3.8 -Cost Activities versus Product Factors
of Printing Labels

3.9 —Product Factors and Associated Ranges

3.10- Examples of Variables and Associated Ranges

3.11-Example of Cost Simulation

xii

19
20
21
29
43
44
47
49
57

63

64

66

67

69
70

70
76
80
82

Table
Table

Table

Table

Table

Table

Table
Table

4.1 -Multiple Variable Regression Analysis
4.2a-Multiple Variable Regression Analysis:
Frc ranges of values doubled
4.2b—Multiple Variable Regression Analysis:
FrC ranges of values increased to equal to
ch ranges of values
4.2c- Multiple Variable Regression Analysis:
ch ranges of values increased to equal to
ch ranges of values

4.2d—Multiple Variable Regression Analysis:

86

86

89

92

Vp ranges of values increased to 5, 500, and 10,000 95

4.3 —Multiple Variable Regression Analysis:

Independent Variables Taken out of the Analysis:

Vp, Frc, Em, and V
4.4 -Inverse of the Correlation Matrix
4.5 -Results of the Regression Analysis:
Tc Digital-Conventional Criterion Variable

xiii

106
108

109

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

LIST OF FIGURES

3.1 —Research Hypotheses Diagram

73

3.2 —Cost Functions of Lithographic versus Digital Printing 76

4.1 -Impact of Frequency of Regulatory Changes on
Total Cost of Label Printing

4.2 -Impact of Frequency of Marketing Changes on
Total Cost of Label Printing

4.3 —Impact of Frequency of Package Engineering
Design Changes on Total Cost of Label Printing

4.4a-Impact of Value of Product-Package System on
Total Cost of Label Printing

4.4b—Impact of Value of Product-Package System on
Total Cost of Label Printing
(Vp value range 5, 500, 10,000)

4.5 -Impact of Number of Brands on
Total Cost of Label Printing

4.6a-Impact of Number of Global Markets on
Total Cost of Label Printing

4.6b—Impact of Number of Global Markets on
Total Cost of Label Printing
(Nm value range 1, 15, 20)

4.7 -Impact of Number of Annual Production Runs on
Total Cost of Label Printing

xiv

87

91

94

97

98

100

102

103

105

Equation 3.1
Equation 3.2
Equation 3.3
Equation 3.4a
Equation 3.4b
Equation 3.5
Equation 3.6
Equation 3.7
Equation 3.8
Equation 3.9
Equation 3.10
Equation 3.11

Equation 3.12a/b —

Equation 4.1

Equation 4.1

LIST OF EQUATIONS

Total Cost of Printing Package Labels
Order Quantity

Printing Cost

Printing Setup Cost Constant (Conventional)
Printing Setup Cost Constant (Digital)
Printing Setup Cost

Printing Cost (Expanded)

Inventory Cost

Average annual inventory volume
Inventory Cost (Expanded)

Stock-out Cost

Service Level

Obsolescence Cost

Standard Error of the Difference Between
Beta Coefficients

The Observed Difference

XV

50
50
51
52
52
52
52
53
53
53
55
55
56

108
108

CHAPTER I

INTRODUCTION

It is only in the past decade-and-a—half that packaging suppliers
have been considered as a strategic component of doing business and
the supply chain (Cole, 1986). Alternative supply chain strategies such
as postponement applied in forms of packaging or labeling can
enhance supply chain competitiveness. Postponement is the delaying
of product differentiation based on form, location, and/or time during
manufacturing and logistics operations. The pharmaceutical industry
for example, depends heavily on the labeling function of packaging.
The timeliness of pharmaceutical labels is especially important for
products under patent protection. Labels also play a very important
role because they convey critical information to the consumers.
Package labeling is also a vital milestone in launching a new drug or
vaccine line, which is going through the initial regulatory approval
process.

With the frequent changes in label regulations for the
pharmaceutical industry, the manufacture/printing of labels becomes
even more critical for increasing supply chain efficiency. Label

obsolescence, inventory issues, and coordination of labeling process

Challenge the industry to provide the right labels at the best time and

place, every time.

GENERAL OVERVIEW

This study examines the cost structure of the supply chain for
packaging label printing. The research is focused on investigating the
label printing activities for pharmaceutical companies to present a total
cost model that can be used to implement postponement strategies.
The goal is to increase supply chain efficiency, maximize patent period
utilization, and minimize total cost of printing labels. The need for this
study has been determined from literature review and interviews with
labeling companies. The model constructed will be validated through
case study interview method.

Labeling postponement and speculation strategies have been
discussed Since 1988 as a cost saving mechanism (Zinn and Bowersox,
1988). Minimizing inventory and maximizing the ability and
effectiveness of responding to customer orders are tools to help realize
the ultimate goal of cost savings. The theory of postponement is a
proven business strategy and has been practiced in the industry and

widely accepted in the academic community.

Zinn and Bowersox (1988) analyzed labeling postponement as a
type of manufacturing postponement and identified the specific costs
that would be impacted. These include: (1) inventory carrying costs;
(2) warehousing; and (3) processing (labeling). Adding to the Zinn and
Bowersox approach, this study identifies the implications of
postponement on the distribution cost for the package labeling
process.

For example, the pharmaceutical industry faces constantly
decreasing operating margins and tapering periods of market
exclusivity between product approval and patent expiration (Hills and
Gugliotti, 2001). For a blockbuster drug (over $1 billion in sales per
year), a one-day delay of the initial launch of a new product costs up
to $3 million (Trombley, 2003). The operating margins are tight and
getting tighter due to Short intervals of patent utilization. The frequent
Changes in the Federal Drug Administration (FDA) regulations also
affect cost of package labeling operations, which may result in delayed
product launch. The regulatory dynamics that are in place contribute
as a critical factor to the timing of the pharmaceutical label production.
Hence, pharmaceutical package labeling has severe cost implications
for the supply chain.

In order to address the impact of package labeling

postponement and costs on the supply chain, this study analyzes cost

drivers for label printing and it identifies product factors influencing
total cost for label printing and the decision to postpone label printing
to the distribution center level. The analysis results in development of
an economic model for implementing postponement strategies. The
cost structure analysis of the pharmaceutical labeling function
contributes significantly to the theory of postponement, promotes
further research, and leads business people to further realize supply

chain cost saving strategies.

BUSINESS PROBLEM

Package labeling is recognized to have great potential for
realizing cost savings in the supply Chain. The challenges faced by the
pharmaceutical industry constitute a pertinent starting point for the
discussion of the business problem. The major impediments that the
industry faces in package labeling are: limited patent period; regular
content Changes, and significant customization. Each are explained in

detail below.

LIMITED PATENT PERIOD

Patent period utilization is critical for pharmaceutical companies.

To effectively utilize the patent period, rapid launch for products upon

approval is key (McNeely, 2001). The final content requirements for
the label are available only after the product is approved (McNeely,
2001). This leaves little time for companies to produce the labels and
launch the product in order to make the most use out of the patent
period (McNeely, 2001). In the case of postponement, the company
chooses to wait until the approval on the final wording for the label is
received. The other strategy is speculation where the company takes a
risk and produces the labels without waiting for the final approval of
wording of the label. The risk is substantial in this case since, if there
are any changes, the whole label batch must be destroyed as Specified
by the FDA regulations and reprinted with the approved content
(McNeely, 2001). Consequently, there are four probable business
scenarios with their respective cost implications as summarized in

Table 1.1 and explained below.

 

 

 

TABLE 1.1
Postponement vs. Speculation for Label Printing
Postpone Speculate
Approval with No Scrap or Rework Scrap and Rework
change(s) Decreased patent period Decreased patent period
Approval as is No Scrap or Rework No Scrap or Rework
Decreased patent period Full patent period utilization

 

 

 

 

 

The advantage of postponing label printing is to avoid the cost of
scrapping and reprinting the entire label batch. On the other hand, the
patent period is decreased compared to speculating when the label
content is approved as is. Speculation is advantageous in the case of
approval as is, but it is the most costly scenario of the four when
approval comes with Changes in label wording. The decision to
postpone or speculate also depends on the type of printing technology
that is employed by the company. The two types of printing processes
utilized for printing on packaging materials are plate or press
processes (conventional), and plateless processes (digital) (Twede and
Selke, 2003). Digital printing technology is an enabler for
postponement, Since setup time is much shorter compared to
conventional printing. With the goal to maximize patent period
utilization, this research investigates the cost components for each

decision and recommends strategies based on total cost savings.

REGULAR CONTENT CHANGES

Label content Change is an on-going activity because of the
obsolescence of label requirements (McNeely, 2001). These changes
are usually based on emerging safety issues. Therefore, the wording of

the labels changes frequently and usually carries grave consequences

for the consumers of the product (McNeely, 2001). This subsequently
affects the demanded turnaround speed of the required package label
supply chain, which could command the application of postponement-
speculation strategies (McNeely, 2001). This could also become an
impediment to patent period utilization through interruptions in
product availability, hence inadequately satisfying the demand. AS a
summary, content changes add a factor (to patent period utilization)
that influences the decision to postpone or speculate. The total cost
model developed in this study, aids in the decision-making process

(i.e. postpone or speculate).

SIGNIFICANT CUSTOMIZATION

The substantial cost of market-specific labeling is a major cost
factor (McNeely, 2001). Customization is especially critical for global
markets with different label requirements, languages, and
measurement units. McNeely (2001) states that digital printing
theoretically presents a cost-effective approach to small-volume
markets. Therefore, in order to serve these markets effectively,
companies Should analyze their supply Chain for implementing
postponement—speculation strategies. This also brings forth the issue

of assessing the value of investing in innovative printing technologies,

such as digital printing, or using providers of such services and
reconfiguring the supply chain composition. Consequently, this study
takes the customization factor into account in developing the tool for

implementing postponement strategies.

BUSINESS PROBLEM SUMMARIZED

These issues portray the importance of designing a model to
analyze the cost structure of the package labeling and produce
quantitative results for implementing postponement strategies.
Postponement applications have been discussed as a global packaging
strategy (Twede, Clarke, and Tait, 2000). Recent research has
indicated that optimization of packaging logistics economies would
justify postponement (Twede, Clarke, and Tait, 2000). Packaging and
labeling postponement have been addressed as postponement
strategies focused on costs associated with transportation,
warehousing, and inventory carrying. (Zinn and Bowersox, 1988). The
challenges affecting labeling (i.e., restrictions and regulations, labeling
for global markets, and patent period utilization for pharmaceutical
companies) have not been addressed using postponement strategies.

Preliminary interviews and brainstorming sessions with The John

Henry Company (Lansing, Michigan) confirmed this Challenge and

revealed the existence of a viable business problem. Existing literature
does not address these problems specifically for label printing.
Therefore, a knowledge gap needs to be addressed and a model
should be developed, which provides a solution to ease the

predicament faced by the pharmaceutical industry.

RESEARCH OBJECTIVES AND QUESTIONS

Package labeling is an important function that has great
potential for realizing cost savings in the supply chain. In the literature
and business practices, postponement strategies are being discussed
and employed as a way to reduce manufacturing cost and increase
efficiency. This study proposes to identify and analyze packaging
postponement strategies and their implications for supply Chain
management. The objectives are to analyze the supply chain function
for package labeling and to formulate a computer-based model that
allows comparison studies by assessing postponement factors based
on inventory costs and strategic benefits. Specifically, this research
addresses the following questions:

. What are the significant cost drivers for package labeling

and how does package labeling impact supply chain costs?

What is the set of product factors that determine the total
cost for label printing and affect the implementation of
postponement strategies in package labeling?

To what degree does each factor affect the total cost for
package labeling and influence the decision to postpone
package labeling to the distribution center level?

What is the relative effect of these factors on
postponement in package labeling?

What types of management strategies can be derived from

the model?

The study aims to use these questions to (1) develop a total cost

model for the package labeling function (2) by defining the parameters

that contribute to the supply chain, and finally (3) to use this model to

suggest supply chain management strategies as they apply to package

labeling operations.

RESEARCH METHOD OVERVIEW

This study proposes to formulate a cost effective supply chain

management tool that addresses the aforementioned issues by

identifying the cost drivers for package labeling. First, product factors

10

that affect the decision to implement postponement strategies are
isolated. The typical value ranges for these product factors are
identified through case study interviews. Next, the effect of each factor
on the total cost of label printing and the decision to postpone is
analyzed, first individually then relative to each other. Finally, the
product factors that have the most significant impact on the total cost
for label printing and the decision to postpone label printing to the
distribution center level is identified.

This study uses the Packaging-Logistics System Integration
concept (Twede and Parsons, 1997) as a starting point in exploring the
package labeling function. Twede and Parsons (1997) indicate that
logistical packaging facilitates all product flow during manufacturing
and distribution. Package labeling is a critical operation supporting the
communication function as part of this integrated system. The
functions of packaging (i.e., protection, utility, and communication)
form the basis of packaging value analysis, which is the foundation for
cost structure analysis (Twede and Parsons, 1997). This concept is
utilized in analyzing the impact of postponement on the total cost of
packagelabefing.

Packaging postponement has been described as the most
interesting example of the value of integration of packaging—logistics

(Twede and Parsons, 1997). Postponement-speculation strategies

11

probe when and where to add value in the supply chain for overall cost
reduction. Since packaging can be a critical cost saving point within
the supply chain, postponement strategies should be studied and cost
structure models should be developed to evaluate the justification of
such strategies.

This study considers all types of postponement (i.e., form and
logistics) (Van Hoek, Commandeur, and V05, 1998). In package
labeling, for example form postponement can occur in situations where
a pharmaceutical manufacturer waits for FDA label specifications
before product launch instead of taking a risk and producing labels by
speculating on the label requirements. Similarly, logistics
postponement occurs when transfer of goods are delayed in the supply
Chain in order to satisfy demand more effectively (Van Hoek,
Commandeur, and V05, 1998). Zinn (1986) concluded that packaging
and logistics postponement—speculation offers the most potential for
realizing cost savings. Generalizations for measuring the extent of
savings should not be based on the predictors of product physical and
demand characteristics (Zinn, 1986). Therefore, this study aims to
identify the factors that formulate the cost model analysis of package
labeling to indicate where postponement strategies are economically

feasible.

12

The Zinn (1986) model of postponement for labeling addresses
the question of postponing the labeling function to the warehouse level
instead of applying labels at the plant. The main cost categories that
Zinn (1986) considered were labeling, inventory, and warehousing.
The model that this study proposes to develop accounts for the
package-label value within the supply chain by focusing on
postponement factors and their effects on inventory costs,
coordination expenses, and strategic benefits. Market characteristics,
such as short product life cycles, benefit tremendously from
postponement.

In order to address these issues and construct the model, cost
drivers in package labeling are identified. The next step consists of
quantifying the effects of postponement on these cost drivers. For this,
a set of product factors that influence total cost of label printing and
the decision to postpone label printing to the distribution center level
must be identified and tested.

To answer these research questions, data is gathered using case
study interviews from alumni and contacts of the School of Packaging
at Michigan State University and the customers of the John Henry
Company, a label printer and manufacturer in Lansing, Michigan. The
case study is a research strategy that comprises an all-encompassing

method incorporating specific approaches to data collection and

13

analysis (Yin, 1994). For the purposes of this study, this method aids
in comprehensively explaining postponement factors for label printing,
and why and how these factors influence the decision to postpone
labeling to the distribution center level. The result is the surfacing of
various manufacturing practices and the reasons behind them. This
helps the results to be placed into context where they may be better

understood.

RESEARCH CONTRIBUTIONS

The underrepresented area of packaging in supply chain
management requires researchers’ attention. The simplest business
oriented reasoning for this stems from the fact that packaging and
logistics postponement-speculation has the most potential for
realization of supply chain savings (Zinn, 1986). In addition, Twede
and Parsons (1997) suggest that the value analysis of a logistical
system's packaging requirements has the potential to dramatically
improve a firm’s profitability. Through postponement-speculation
strategies, firms can realize more profitability. Contemporary printing
technologies can be enablers to postponing label printing to the
distribution center level. However, digital printing technologies present

complications in complying with label content security measures, as

14

mentioned previously. Therefore, with the emergence of digital
printing technologies the package label regulations can only be
expected to intensify and the issue of label obsolescence can be
significant. In addition, immediate apprehension of significant cost
savings can be anticipated by optimizing the cost structure of label
printing. Consequently, this study develops an evaluative tool to
produce a cost structure analysis of the package label supply Chain.

Package labeling is also a critical function within the supply chain
in its role of providing accurate information. This is especially
important in the pharmaceutical industry. The cost model of the
package labeling function is quite complicated. It includes a setup cost,
which generally justifies high quantity production. On the other hand,
frequent changes in label requirements push manufacturers toward
manufacturing small quantities. This needs to be investigated in order
to bring solutions to these problems and to come up with optimal
management strategies.

This study analyzes potential cost savings and recommends
postponement strategies for package labeling operations. The model
developed also forms a basis for Similar cost analysis tools in other
areas of supply Chain management. Overall, this research paves the
way for significant cost savings and points to further research on the

critical function of supply chain management (i.e., package labeling).

15

CHAPTER II

LITERATURE REVIEW

This chapter is divided into four sections each corresponding to
the research streams on which this study is based. The first section
provides an overview of the current printing technologies for labels.
The second section reviews the postponement-speculation research
field. The third section focuses on the activity-based costing model to
establish a reasonably accurate way to track the cost items of label
printing. In the fourth section, the total cost concept is introduced and
its applications are discussed. The last section identifies the gap in
literature, discusses the ways in which this study is based on these
four areas of research, and presents the justifications as to how this

research addresses these issues.

PRINTING TECHNOLOGIES

There are two types of printing processes utilized for printing on
packaging materials: plate or press processes, and plateless processes
(Twede and Selke, 2003). Plateless process is more recent and
produced using digital imaging technology and is referred to as digital

printing technology in this study. The plate printing processes used

16

most for packaging include flexography, rotogravure, and offset
lithography (Twede and Selke, 2003). For the purposes of this study,
these are referred to as conventional printing methods. The two
methods (i.e., digital/plateless printing and conventional/plate
printing) present different cost structures and bring different solutions
to the printing and converting industry.

One important characteristic of plateless/digital printing includes
fast label turnaround, which is due to replacement of plates and films
used in plate printing process with electronic data (Clarke, 2001). This
also endows digital printing with minimal setup times, thus decreasing
the associated fixed costs (Emden, 2001-2003). Images and text are
prepared and proofed electronically in digital printing (Clarke, 2001).
This presents some challenges in terms of securing the integrity of the
label content that has been proofed and approved for printing (Emden,
2001-2003). Since data can be modified in real time, the physical
forces that are used in plate printing to ensure integrity do not apply
to digital circumstances most of the time.

Labeling regulations can be expected to tighten through
emergence of digital printing technologies and the efficiencies that this
technology brings. With traditional printing methods, such as
lithographic printing, plates are approved for each label job. Digital

printing, on the other hand, has the capability of modifying label

17

content and design in real time and on-demand, which endangers the
viability of security measures. Such security measures involve absolute
containment of the label content, preventing it from being modified
once it has been approved. In the case of digital printing, approved
label content is in the form of an electronic file, hence it is easier and
faster to modify.

The opportunity for better information sharing is increased with
digital printing through electronic data storage (Clarke, 2001). In
addition, the quick turnaround and minimal setup costs associated with
digital printing makes delaying the label printing activity possible,
hence allowing postponement strategies (Clarke, 2001). With digital
printing, costs are justifiable for small quantity orders (i.e., customized
wine bottle labels for a specific party) (Emden, 2001-2003). From a
cost perspective, digital printing has lower fixed costs but higher
variable costs above a certain quantity of labels when compared to
plate printing (Emden, 2001-2003 and Bence, 2001).

Falkman (2001) presented five primary reasons for the
packaging industry to switch to digital printing. Table 2.1 below
summarizes these reasons. First, there could be substantial cost
savings for short-run printing jobs due to elimination of making the
print plates. Second, customization of graphics or sequential

numbering is possible. Third, digital printing also provides fast

18

turnaround times, at least for short-run jobs. Fourth, product samples
and prototype packages could be made through digital printing, since
they are one-of-a-kind. Lastly, digital press is used for proofing, even
if the job is transferred to a conventional flexo/litho press for the full
run after approvals are received.

TABLE 2.1

Reasons for Digital Printing in the Industry

 

1. Cost savings for short-run jobs due to minimal investment of
time or money (print plates are not needed).

 

. Customization for graphics or for sequential numbering.

 

Fast turnaround times, especially for short-run jobs.

 

wa

. Supporting product samples and prototype packages well, due to
one-of-a-kind nature.

 

 

5. Providing proof process for flexo/litho jobs.

 

 

Digital printing tends to be more cost effective in short
production runs and plate printing is more likely to be cost effective in
long production runs. Multiple solutions presents challenges for
managers in making decisions with cost effective results. The total cost
model constructed in this research accounts for both types of printing
technologies and suggests printing solutions pertaining to different

cost structures.

19

 

POSTPONEMENT-SPECULATION STRATEGIES

Postponement is the delaying of product differentiation based on
form, location, and time during manufacturing and logistics operations.
For example, Hewlett-Packard (HP) delays the individual packaging of
its DeskJet printers until the products arrive at the distribution centers
(Howard, 1991). The types of postponement discussed in literature are
time (logistics) and form (manufacturing) postponement (Pagh and
Cooper, 1998). Bowersox and Zinn (1988) introduced four
manufacturing postponement strategies including labeling, packaging,
manufacturing, and assembly, in addition to time postponement
(through centralized location) under the time (logistics) type. These
are summarized in Table 2.2. Both labeling and packaging are
analyzed as forms of manufacturing postponement in the Bowersox
and Zinn study. The present study however, looks at implementing
postponement strategies, regardless of the type (i.e., time and
manufacturing) in addressing the business problem identified.

TABLE 2.2

Types of Postponement

 

 

 

 

Manufacturing (Form) Postponement Logistics (Time) Postponement
Labeling, Packaging, Manufacturing, Centralized Location
Assembly

 

20

 

Producing goods based on forecasting the demand is referred to
as speculation. The aim is to gain economy of scale in manufacturing
and logistics and reduce stock-outs (Pagh and Cooper, 1998). The
postponement-speculation matrix by Pagh and Cooper (1998)
presents the four possibilities as summarized in Table 2.3. These
include: speculation in logistics and manufacturing, postponement in
logistics and manufacturing, postponement in logistics and speculation
in manufacturing, and speculation in logistics and postponement in
manufacturing.

TABLE 2.3

Postponement-Speculation Matrix

 

 

 

 

 

 

 

M L o G I s T I c s

i: Speculation Postponement
u Speculation The full speculation The logistics

it (make to inventory) strategy postponement
g strategy

: Postponement The manufacturing The full

in (make to order) postponement postponement
G strategy strategy

 

 

 

 

Source: Pagh and Cooper, 1998

Examples of postponement strategies in business logistics
practice have been recently revealed (Bowersox and Closs, 1996). This
indicates an enduring need in research and business practice to

investigate possible venues for implementation of these strategies.

21

 

HP’s success in employing postponement-speculation strategies in the
early 19905 exhibits the potential savings that still exist for the supply
Chain.

Identifying responsible parties in the implementation of
postponement strategies remains crucial. For example, HP’s case of
employing product differentiation and packaging postponement
significantly shifted responsibilities from the manufacturing plants to
the distribution centers (Howard, 1991). This helped HP realize
tremendous cost savings over a short period of time (Howard, 1991),
which once again indicates the potential that exist for cost savings and
improved manufacturing practices.

In addition, with a direct link to e-Commerce, the adoption of
digital printing technologies is increasing, which allows easier
implementation of postponement strategies (Clarke, 2001).
Similarly, industry interest for increased computer, Internet, mobile
phone linked data generation (Schoenmakers, 2001) and supply Chain
Visibility drives more and more companies to consider postponement
strategies. Furthermore, the TINA (There Is No Alternative) trends
identified by DLF/Delfi in 1998 for Nordic trade apply to global trade as
well. These trends include information technology, efficiency,
consumer awareness, intelligent marketing, and consumer adapted

retailing concepts (Olsmats, 2000); and also aids in forming the

22

infrastructure to implement postponement strategies. Therefore, the
current study of analysis and implementation of postponement
strategies in package labeling is quite timely.

The literature highlights postponement strategy and its effect on
supply chain efficiencies and cost savings in various industries. One
good example is the HP DeskJet printer case. HP was able to cultivate
the best features of the postponement—speculation strategies and
transform the long-established lines of responsibility between
manufacturing and warehousing toward a very successful global
implementation of product differentiation and packaging postponement
(Howard, 1991).

The key to achieving high levels of product customization is
postponement of product differentiation until later in the supply chain
network (Feitzinger and Lee, 1997). Increasing flexibility and lowering
costs are two motivating goals for a company to standardize
components for different products or to postpone the assembly of
differentiated components until later in the manufacturing process
(Feitzinger and Lee, 1997). Lee examined HP’s process redesign for
their printers that were being manufactured in the US. and shipped
globally (Lee, Billington, and Carter, 1993). By postponing the
differentiated components, namely the power supply and the user’s

manual, until the product arrives at the distribution center, HP realized

23

tremendous cost savings resulting from reduced inventory and
improved fill rates (Johnson and Anderson, 2000).

Building on this, Johnson and Anderson (2000) developed a
model that analyzes the value of postponement. Johnson and
Anderson (2000) explored factors such as inventory policy, forecast
uncertainty, product variety, product mix, and postponement premium
and then introduced a model which evaluated the impact of
postponement on introducing product platforms. Similarly, this study
identifies and examines the effects of product factors on the total cost
for package labeling and the decision to postpone.

Zinn and Bowersox (1988) identified a methodology for
postponement justification by introducing five types of postponement
and developing cost models for each type. Among the types of
postponement, the researchers included labeling and packaging, and
identified cost categories associated with each. The cost categories for
these two types of postponement were identified as: (1) inventory
carrying costs; (2) warehousing; (3) processing (labeling or packaging
respectively); and (4) transportation (packaging only) (Zinn and
Bowersox, 1988). Suggestions for postponement justification produced
by this research seem to have re-ignited the discussion on

postponement among other researchers.

24

 

Lee and Tang (1997) discussed the concept of reorganizing the
production process so that the product differentiation point would be
delayed as long as possible, hence delayed product differentiation. The
researchers came up with a model for analyzing costs and benefits of
postponement through product/process redesign. This model
accounted for factors such as processing cost, inventory cost at work-
in-progress stages, lead times, design cost, etc. (Lee and Tang, 1997).
The model enables businesses to redesign their production processes
so that the complexity of manufacturing process is reduced and the
flexibility is increased (Lee and Tang, 1997).

The inventory holding cost analysis as presented by Waller, et al.
stated that it is more expensive to hold finished products than it is to
hold materials. This is primarily due to the higher labor and material
costs associated with finished product inventory as compared to
material inventory (Waller, Dabholkar, and Gentry, 2000). The
researchers concluded that the marginal benefit of postponing a
process at the beginning of the supply chain would be less than that
toward the end of the supply chain (Waller, Dabholkar, and Gentry,
2000). Labeling is positioned relatively towards the end of the
manufacturing process. Therefore, this reinforces the focus in this
study on postponement of package labeling due to its positioning in

the supply chain.

25

Battezzati and Magnani (2000) confirm the classification of
postponement into two types: logistics and manufacturing. In their
research, they stated that logistics postponement is applicable in all
situations with the most realized benefits when the distribution
component incorporates product customization (Battezzati and
Magnani, 2000). Package labeling can be considered part of the
distribution process. For example, frequently changing Food and Drug
Administration (FDA) label requirements has a major impact on
pharmaceutical products, especially recently launched drugs waiting
for regulatory approval. Manufacturing postponement on the other
hand, results in efficiency when the process is redesigned to be
modular and product differentiation is delayed to the last steps in the
production process. Consequently, this study satisfies an existing
knowledge gap in the literature by developing a total cost model to
address these issues for pharmaceutical companies.

The benefits of postponement become even more visible in
multiple markets. For example, European markets have been
investigated for postponement justification and it has been concluded
that assembly postponement strategies yield considerable benefits
(Foster, 2000). Due to the multiple markets existing within Europe,
postponement strategies bring in considerable savings even when they

are applied at the package labeling and documentation stage (Foster,

26

wa

 

2000). The savings mainly result from not having to stock up on the
same product labeled/packaged and documented for countries in
different languages (Foster, 2000). This documents the need to study
global markets in developing the total cost model for implementing
postponement strategies. Consequently, in this study the complexities
of global markets are accounted for. The model develops solutions for
pharmaceutical companies to address the business problem noted
previously.

Twede, Clarke, and Tait (2000) discuss postponement
applications as a global packaging strategy. Factors that justify
packaging postponement are briefly discussed, such as characteristics
of the product, market/demand structure, and manufacturing/logistics
setup (Twede, Clarke, and Tait, 2000). The authors conclude that
optimization of packaging and logistics economies of scale would
justify postponement (Twede, Clarke, and Tait, 2000). Therefore, it is
important at this time to create a model, which analyzes cost
structures for implementation of postponement strategies in the
packaging logistics function.

Postponement of packaging can reduce inventory and
transportation costs, especially when the supply chain has a
complicated structure (Richardson, 2001). More specifically, since

package labels are being differentiated more and suppliers are

27

’-

applying more labels on packages (Richardson, 2001), postponement
deserves further attention at this stage of the supply chain. One way
of measuring this could include looking at the number of brands a
company has. Thus, the present study considers the number of brands
as a product factor that affects the total cost model for postponement
decisions.

The literature review by Pagh and Cooper (1998) produced the
postponement—speculation matrix for supply chain (see Table 2.3).
The research emphasizes the importance of the integration of logistics
and manufacturing activities and the role of postponement—speculation
as an emerging strategy in effective supply Chain management (Pagh
and Cooper, 1998). The researchers also indicate that limited
considerable effort has been made to bring postponement-speculation
theory into practice for managerial decision-making. Focusing on the
manufacturing plant level to end consumer, their research develops a
diagnostic and normative framework for deciding on postponement-
speculation strategies. The decision factors are categorized into three
groups, as tabulated in Table 2.4, and are classified as follows (Pagh
and Cooper 1998):

1. Product — Emphasis is on the life cycle and value of the

product similar to Zinn’s (1986) analysis.

28

 

2. Market and Demand — Focus is primarily on the relative
delivery frequency and demand uncertainty.

3. Manufacturing and Logistics System — Highlighted area is the
economies of scale or special knowledge needed in the
processes.

TABLE 2.4

Postponement-Speculation Decision Factors

 

 

 

 

Product Market and Demand Manufacturing and
Logistics System
Life Cycle Relative Delivery Economies of Scale
Frequency
Product Value Demand Uncertainty Special Process Knowledge

 

 

 

Bowersox and Closs (1996) formulate postponement as time or
logistics and form or manufacturing, consistent with previous
literature. Some examples of manufacturing postponement include
mixing of different colors of paint at the retail level based on direct
demand from the consumers, blending octane grades at the retail
pumps (Sunoco), and postponing the installation of market
differentiated parts of HP DeskJet printers (e.g. parts dependent on
voltage) to the distribution center level. All these cases are indicative
of the business potential underlying postponement.

In a literature review by Van Hoek (2001), the pharmaceutical
industry is referred as a process industry. Therefore, the processing

cycle times might last longer than the customer order lead time, and

29

 

 

the process cannot be decoupled at an intermediate stage (Van Hoek,
2001). The operating Characteristics of the industry determine the
feasibility of postponement types (Van Hoek, 2001). Package labeling
constitutes a complicated manufacturing step and consequently needs
researchers’ attention in analyzing the problems associated with it. In
support of this statement, Van Hoek (2001) refers to a conversation
with Bucklin; who is the author of the founding article on
postponement theory (Bucklin, 1965); from which he quotes
“Postponement was introduced in the 19605 but look at the entire
speculative inventory that is still stored in the channel” (Bucklin,
1965). Therefore, need for further research in this area is ample.
Following what has been practiced and researched so far, the
objective of this study is to bring together a comprehensive evaluative
tool that performs cost structure analysis for justification of
postponement strategies in package labeling. As explained with
numerous business cases and proven studies, the need for developing

this economic model for package labeling postponement is evident.

30

 

ACTIVITY- BASED COSTING

In order to accurately devise an economic model, first the cost
structure of the labeling industry needs to be understood.
Consequently, both indirect and direct costs should be taken into
consideration. For the purposes of modeling or decision-making, it is
necessary to develop cost relationships of direct and indirect costs,
rather than estimates (Shapiro, 1999). Activity—Based Costing (ABC) is
widely used in practice and seeks to allocate indirect costs (i.e.
activities, to cost objects such as service, product, and customer
costs) (Shapiro, 1999). In addition, ABC is referred to in the literature
as “being approximately right” versus “being precisely wrong” (Ernst &
Young LLP, 1994).

While the allocation of direct costs engages obvious drivers, such
as machine hours for machine cost, indirect costs are not that
straightforward (Shapiro, 1999). Shapiro indicates that indirect cost
drivers may be volumetric, such as a cost driver for a receiving
department that can be measured by the number of parts handled by
the department during the planning period (1999). Or they may be
non-volumetric; such as a cost driver for machine setup costs that
equals the number of times a machine is setup during the planning

period (Shapiro, 1999). Wiersema (1995) refers to the fact that

31

 

traditional accounting tends to average out all costs, direct and
indirect, and displays unit-driven results. However, batch-driven
activities, such as label printing, include a setup component, which
varies regardless of the number/volume of units produced. Therefore,
it is crucial to incorporate such factors as they may drastically change
the total cost of performing a particular job. As explained in the next
chapter, allocation of the indirect cost drivers are very important and
used to arrive at some of the differentiating factors for postponement.
It is also noted that the connection between ABC and supply
Chain optimization models is important but also complicated (Shapiro,
1999). In general, ABC starts out by mapping indirect costs in the
company general ledger into cost categories (in this case, supply
chain) with associated cost relationships, drivers, and resources (Ernst
& Young LLP, 1994). With supply chain cost structure in focus, Shapiro
(1999) indicates that the cost driver corresponds to a resource if its
availability is limited. Shapiro also refers to the difficulty in interpreting
accounting systems (i.e. calibrating costs across facilities due to use of
multiple accounting systems) in modeling the supply chain (1999). The
different cost structures used for digital and conventional printing need
to be unveiled using activity-based costing; in order to identify the real

cost drivers accurately.

32

 

One major contributor to the difference in cost structures
between conventional and digital printing is setup costs (Emden, 2001-
2003). Since setup costs tend to be calculated as indirect costs
(Shapiro, 1999), ABC will be the most useful method in modeling the
cost structures associated with digital and conventional printing. In
addition, as explained in the next Chapter, this research utilizes the
case study research approach, which alleviates the aforementioned
problem. The case study method provides comprehensive insight into
the cost variables and factors involved in managerial decision-making.

Since ABC and optimization modeling are recognized to play
complementary roles in identifying costs and cost relationships for
supply Chain decision-making, this study utilizes ABC principles and
methodologies in identifying costs and cost relationships for package
labeling. This assists in devising a cost model that properly accounts

for the indirect costs for label printing, such as setup costs.

33

 

TOTAL COST APPROACH

This research takes two supply chain management concepts into
consideration: The total cost approach and postponement-speculation.
Packaging is an integral part of the supply Chain and therefore needs
to be studied under theories applicable to supply chain management.
The total cost concept is introduced, reviewed, and evaluated for its
effects on postponement.

It is widely accepted that the concept of total cost logistics was
first introduced by Lewis, Culliton, and Steele in 1956 when they
justified an increase in the cost of air transport through the reduced
total cost of the logistics chain (Bowersox and Closs, 1996; Lewis,
Culliton, and Steele, 1956). The principle of total cost logistics states
that instead of trying to reduce individual cost components in the
logistics network, it is more effective to devise strategies that reduce
the total cost of the logistics system. Decreasing the cost of one
component such as plant to warehouse transportation does not
necessarily imply that reduction in the total logistics cost will be
realized. Through total cost approach, expenditures can be justified
where needed, such as delivering goods fresh to consumers, while

cutting back on activities that do not add value to the supply chain in

34

 

an attempt to reduce the total cost of logistics (Bowersox and Closs,
1996).

In order to devise a model to be used in the implementation of
postponement strategies, it is necessary to carefully examine the cost
structure of the subject matter (i.e., package labeling) and maintain a
total cost approach. Consequently in this study, an objective is to
minimize the cost of doing business, the technique used is
postponement, and the tool is an evaluative model designed with a
total cost perspective.

Package labeling requires extensive coordination between the
other supply chain activities, and it is also this coordination that
savings can be realized. The supply chain is an integrated system, and
therefore the efficiency and performance level of one activity will be
affected by the other components. Bowersox and Closs (1996) support
this in examining total cost analysis within the framework of an
integrated logistics system. An appropriate example is the Hewlett—
Packard Company’s DeskJet printers case study , which shows lower
total costs even though some of the cost items escalate (Howard,
1991)

Zinn (1986) investigated postponement strategies in business
logistics by focusing on cost trade-offs within the objective of

minimizing total cost of the business logistics. The study used the six

35

If

:g-l um».- Y.

 

cost categories identified by Lambert and Stock (1982): 1)
transportation costs, 2) warehousing costs, 3) inventory carrying
costs, 4) order processing and information costs, 5) production lot
quantity costs, and 6) cost of lost sales (Zinn, 1986). The cost
categories for labeling postponement have been identified as labeling,
inventory, and warehousing (Zinn, 1986). Adding to the Zinn model of
postponement, this study includes obsolescence cost of labels
including time/expiration, regulatory/safety changes, marketing based
changes (e.g. graphic Changes on package) package engineering
changes, and product line changes. In addition to obsolescence costs,
the cost of elapsed time between approval and launching of patented
products and setup costs for small volume or short run products also
need to be considered in the model. All these cost factors make up the
total cost structure of package labeling, and therefore need to be
considered in developing the total cost model for implementing

postponement strategies.

ANALYSIS OF PACKAGE LABELING

In this study digital printing is reviewed as an enabling
technology to postponement of label printing. Compared to

conventional (plate printing), electrographic (digital or plateless)

36

 

printing has minimal setup time and costs (Emden, 2001-2003).
Delaying the printing activity is likely to be considered more by
management since plate printing requires additional setup time.
Second, the cost per unit produced by digital printing does not vary
significantly with increased output to justify long production lines so
work can be interrupted or performed at different times (i.e.
conventional printing is more economical when compared with digital
in long production run) (Emden, 2001-2003). Additionally, since setup
activity is minimal compared to conventional printing, customization of
labels is easier especially in multiple language production or even
when presenting different information (Larson, 2000). Therefore,
micro-segmentation or order-driven printing is made possible (Larson,
2000). Finally, the fact that digital printing techniques enable quick
product launching to market (Larson, 2000) assists in delayed product
differentiation (i.e. postponement).

So far, researchers have not addressed this aspect of enabling
technologies in the area of package labeling. This study investigates an
economic model for label printing and its effects on supply chain
management. The research seeks economic justification for
implementing postponement strategies by accounting for the effects of

enabling technologies on label printing operations.

37

lit—1*...._._____u_._h._

Even though the concept of postponement is not new to the
supply chain management literature, the impact of labeling
postponement on the supply Chain has not been specifically addressed.
Only one cost-benefit analysis of labeling postponement has been
performed (Zinn, 1986). However this model did not encompass
regulatory (government) or strategy-based factors. Even though
value-based factors were included in the model, value of product-
package system and package engineering design costs were omitted
from the equation. In addition, regulatory factors play an increasingly
important role in the pharmaceutical industry, which is highlighted in
this research. Hence, there is a gap in literature and a need to
comprehensively analyze cost function of package labeling and its
effects on the supply Chain.

The cost structure of label printing should be investigated from
an Activity-Based Costing (ABC) point of view with a total cost
perspective to understand business decisions that are driven by
dynamics based on three categories, namely government and market,
strategy, and value. These categories control the factors that have a
significant impact on the total cost for printing labels. In addition, it is
evident from the examples provided above that in the literature and
business practice, postponement strategies are still being explored as

a way to reduce manufacturing costs and gain supply chain

38

efficiencies. There is still a tremendous potential for cost savings in
postponement strategies and therefore, this research proposes to
uncover when it is cost effective to implement postponement
strategies in label printing, specifically for pharmaceutical companies.
Another area where current literature falls short is in the approach to
postponement strategies through ABC. This research utilizes the ABC
model to deliver a total cost model for package labeling. Many cost
items that may be hidden or inaccurately calculated can be accounted
for in an “approximately right” manner.

In order to address these needs, this study aims to identify
factors that are driven by government and market, strategy, and value
based causes. Next, the research proposes to formulate an economic
model for package label printing to implement postponement
strategies. This model paves the way to a better understanding of the
label printing operation and its position in the supply chain. The study
effectively assesses supply chain management strategies in reducing
total cost and accurately pinpoints areas that have the greatest impact

on total cost for label printing and the decision to postpone.

39

CHAPTER III

MODEL DEVELOPMENT AND RESEARCH DESIGN

The model derived from the research questions reiterated below,

eventually assists in quantifying the effects of package labeling

postponement on the whole supply Chain and provides management

suggestions regarding use of postponement. Therefore, the following

questions are addressed in this section:

What are the significant cost drivers for package labeling and how
does package labeling impact supply chain costs?

What is the set of product factors that involve cost components that
make up the total cost for label printing and affect the
implementation of postponement strategies in package labeling?

To what degree does each factor affect the total cost for package
labeling and influence the decision to postpone package labeling to
the distribution center level?

What is the relative effect of these product factors and the
individual cost components on postponement in package labeling?
What types of management strategies can be derived from the
model?

In order to craft answers to these questions, the following steps

are taken:

40

Conceptualize the product and cost factors that generate the model
forpnnfinglabeb;

Establish the relationship between these product factors and the
total cost of printing package labels;

Define relevant business environment by analyzing different
business scenarios that is established by using typical values for the
product factors;

Determine the correlation between cost factors and the decision to
postpone package labeling to the distribution center level, which in
turn generates the cost model for printing labels.

The total cost simulation model is developed using Microsoft®
Excel® and multiple variable regression analysis is conducted to assess
the statistical significance of the independent variables.

This chapter is divided into four sections. The section titled
Model and Theory Development introduces the research model and
develops the theory for this study including cost and product factors
that are included in the model of package label printing. The second
section describes the model structure, which explains the relationships
between the product and cost factors for package labeling. Research
hypotheses are presented in the following section. The last section
describes the method to be used and the overall design of the

research.

41

MODEL AND THEORY DEVELOPMENT

In developing the model, the cost structure of package label
printing is examined. A series of preliminary and exploratory
interviews were conducted with alumni and contacts of the School of
Packaging at Michigan State University and the customers of the John
Henry Company to uncover factors that have a significant impact on
the total cost of printing labels. As explained in Chapters I and II the
purpose of the study is best served by restricting the research to the
pharmaceutical industry. This provides a focused approach for
addressing the business problems specified in chapter I (i.e., limited
patent period; regular content changes, and significant customization).

A total of seven features are identified for this business segment
that stem from three categories; namely government and market,
strategy, and value. These features are summarized in Table 3.1 with
corresponding categories. The government enforces regulatory
changes based on new medications and their implications on the
public. The seasonality of product also affects inventory levels, which
in turn affect total cost for printing labels. These features therefore
represent outside forces that the company has little or no control over.
In order to maximize business, companies employ strategic

approaches on package labels such as monitoring market demand and

42

changing label graphics accordingly, producing the same product
under different brands, and expanding their business into global
markets. Lastly, firms utilize value-based features such as package
engineering design to lengthen the shelf life of a product. Value of any
product-package system ultimately determines the price of the product
that leads to the profitability of the product line.

TABLE 3.1

Features Impacting Cost of Printing Labels

 

 

 

Government & Market Based Strategy Based Value Based
- Regulatory Changes - Marketing based - Package Engineering
(government based) Changes Design Changes
- Seasonality of Product - Number of Brands - Value of Product-
(market based) - Number of Global Package System
Markets

 

 

 

 

As explained in Chapters I and II, the case study interview
method is used to setup the total cost model, identify the
representative minimum, mode, and maximum ranges of values for
each variable included in the model, and the business environment.
The method is discussed in Chapter II and in the methodology and
research design section in this chapter. A total of seven product factors
affecting package-label printing are derived from the features that are
introduced above, tabulated in Table 3.2, and explained in detail in the

following section.

43

TABLE 3.2

Product Factors Affecting Cost of Printing Labels

 

 

Regulatory Changes —> Frequency of Regulatory Changes (Fm)
Marketing based changes —> Frequency of Marketing Changes (Fm)
Package Engineering Changes —> Frequency of Packaging Changes (ch)
Value of Product-Package System —> Value of Product-Package System (VP)

Number of Brands —> Number of Brands (Nb)
Number of Global Markets —> Number of Global Markets (N...)
Seasonality of Product —> Number of Annual Production Runs (NW)

 

A total of seven product factors affecting the total cost of
printing labels are identified. Frequency of regulatory changes (Fm) is a
government driven effect. Frequency of marketing based changes
(ch) is the consequence of strategy-based decisions. Frequency of
package engineering design changes (fix) is the result of value-based
decisions.

Similarly, value of product-package system (Vp) is the
consequence of value based business features. Number of brands (Nb)
and global markets (Nm) result from strategy based decisions. Finally,
the number of annual production runs (Npr) is the outcome of the
seasonality of a product and additional market based features. The
seven product factors are explained in detail under the Product Factors
subheading in this Chapter. The cost factors that make up the total

cost of printing labels are explained next.

44

 

COST FACTORS

From a business perspective, the motivation to postpone is to
maximize profits primarily through inventory reduction. In a general
sense, profits equal revenues minus total cost. For the purposes of this
study, revenues are assumed constant. However, it is also important
to recognize the fact that postponement also has effects on revenue.
For example, by utilizing postponement strategies, companies are able
to customize certain products according to customers’ needs, and
therefore, have the customer pay more for the product than a more
generic counterpart.

From the pharmaceutical industry point of view, the revenue
gain will be realized through faster launch of patented products once
approvals are received. Postponement-speculation has diverse effects
on the speed of launch of a particular product. For example, a
company that postpones label printing until product approvals are
received may actually shorten the product’s time to market. On the
other hand, a costly situation may occur when labels are printed prior
to receiving the approvals with a speculative approach on the label
requirements and then to have the labels and/or inserts re-printed

with accurate information.

45

Another business scenario is to print the labels with speculative
wording and actually end up gaining time in launching the product if
the approval is received without any changes in label wording. There
are situations where pharmaceutical companies make such decisions
based on experience. When materials are printed before approval, it is
described as “at risk,” meaning that if there are any Changes in the
wording for the labeling, the printed materials have to be destroyed
(McNeely, 2001). In addition, McNeely (2001) notes that it is
sometimes advisable to print the inserts “at risk,” since it becomes a
rate—limiting step in the launch process. However, there are some
cases where not taking such risks may also depend on the company
having digital printing capabilities that boast minimal setup time, and
this alone may justify the costs associated with labeling postponement.

In this research, revenue is assumed to be constant, and
therefore maximizing profit will depend solely on minimized total cost.
At the same time, the study recognizes the importance of analyzing
the effects of postponement on revenues, both from a purely research
contribution and from a business perspective. Therefore, the impact of
postponement strategies on revenue is discussed in the last chapter.

In the traditional cost accounting method, the cost elements for
the labeling function can be grouped into two cost categories. These

are fixed and variable costs. Fixed costs include personnel, utilities,

46

depreciation (equipment), service (equipment), and depreciation
(facility). Variable costs include substrate, colorant, equipment
maintenance, and inventory carrying/warehousing. Fixed costs are
normalized based on calculated totals on a per hour basis (QPR,
2002). The total cost then is the sum of fixed and variable costs. The
traditional cost model of label printing is summarized in Table 3.3.

The traditional cost accounting model, however, does not
accurately represent the indirect cost items involved in label printing.
To compute these unaccounted sources of variance, the Activity-Based
Costing (ABC) methodology should be followed, as explained in

Chapter II.

TABLE 3.3
Traditional Cost Model of Label Printing

 

Fixed Costs

 

Personnel/Labor ($50,000/person annual, 30 people for litho, 15 for digital)

 

Utilities ($10,000/year)

 

Depreciation (Equipment)

 

Service (Equipment, $5,000/month digital, $10,000/month litho)

 

Depreciation (Facility)

 

Variable Costs

 

Substrate ($1 per million square inches (msi), 20,000 8 1/2 x 11 sheets per hour)

 

Colorant ($300 digital, $40 litho)

 

Equipment Maintenance/Cost of parts

 

 

Inventory Carrying/Warehousng

 

 

Source: Emden, 2001 -2003

ABC advocates Claim that activities consume resources, and re-

doing or re—working unacceptable work is an activity (Dickeson, 2000).

47

w

Accounting for this is important, especially, when manufacturers print
labels “at risk” and have to re-do the work due to regulatory Changes
imposed upon approval of a new drug. In such cases, ABC principles
need to be followed to properly account for the cost. Traditional
costing methods do not account for activities such as setup in batch
processes (Wiersema, 1995). Printing labels is a batch process. There
is a setup activity required to print a certain label for a particular
product, which stays constant whether the quantity of labels to be
produced is 10 or 10,000 linear feet. This further suggests that the
problem at hand needs to be approached utilizing ABC principles.
Following the Wiersema (1995) ABC model the activities for label
printing can be regrouped as seen in Table 3.4. The process of
implementing Activity Based Management (ABM) is introduced by first
grouping the activities into two categories, namely Support Activities
and Operating Activities. The next step involves assembling individual
activities under these two activity categories (Wiersema, 1995).
Hence, the cost category of operating activities is divided into direct,
discretionary, and special requirements activities (Wiersema, 1995).
Note that compared to information in Table 3.3, this structure
identifies additional costs relating to label printing, such as setup

costs, and segregates the cost items as they relate to activities.

48

F."

According to the foremost ABC principle, activities consume resources,
which this study follows. Table 3.4 presents these activities.

Support activity costs do not vary with the seven identified
product factors and therefore are assumed not to have any effect on
postponement decisions. Consequently, the main focus of this study is
operational costs (i.e., costs that are driven by operating activities).
This way, hidden costs associated with printing setup, inventory
carrying, handling obsolete inventory, and more are revealed.
Eventually, the effects of the seven product factors on the total cost of
label printing and the decision to postpone labeling to the warehouse
level are analyzed.

TABLE 3.4
Activity-Based Cost Model of Label Printing

A. Support Activities
. Personnel/Labor
. Dggreciation (Equipment)
. Service (Eguipment)
. Utilities
. Depreciation (Facility)
B. Operatig Activities

1. Direct Activities
A. PrintiniSetup
B. Printing Run
1. Substrate Usage Rate
2. Colorant Usage Rate
3. Equipment Parts Replacement Rate
4. Inventory Carrying/Warehousing
2. Discretionary Activities
A. Abnormal Scrap/Rework
3. Special Requirements Activities
A. Special Customer Requests (such as expedited turnaround, customized
labels)

 

 

 

 

 

 

UTAOJNI—l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

49

Regarding government and market, strategy, and value drivers
for the product factors, the total cost of label printing involves printing
production run (PR), inventory (INV), stock-out (STD), and
obsolescence (OBS) costs. The percentage difference between
projected and budgeted costs is incorporated into the case study
interviews to incorporate an error percentage term into the total cost
calculation. The error term of up to 10% is then used to recalculate
the dependent variable (total cost). The data gathered through the
case study interviews are used to establish the representative ranges

of values for the cost factors and the variables used to calculate them.

Tc = PR+INV+STO+OBS (3.1)

Functions for each of the cost factors are written out in terms of
the independent variables product factors and other factors such as
average annual volume (V), service level objective (SL), stock-out cost
per day (STOpd), and average annual order quantity (0Q). The volume
and service level objective determine the production cycle.

STOpd = Stock-out cost per day

OQ = Order quantity

V = Average annual volume

0Q = V / (Npr /2) (3-2)

50

 

Calculation for Printing Run Cost

Printing costs (PR) include technology costs that are associated
with a particular technology adoption (i.e., digital versus litho) in
addition to the volume of labels printed. Therefore, the activities that
drive this cost category are all the direct activities (i.e., printing setup,
substrate usage, colorant usage, and equipment parts under the
printing run activity item). Table 3.4 summarizes these cost activities.
The two product factors that directly impact the printing setup costs
are number of global markets (N...) and number of brands (Nb). At this
point, it is beneficial to reiterate that the support activities are not
impacted by the seven product factors; hence they are not taken into
account in the total cost model. The equation below demonstrates the

calculation of printing run costs.

PR = P5 + (CONSTPR X V) (3.3)
PS = Printing setup cost

CONSTpR = Printing Run Cost Constant

CONSTpR represents the annual expenses on substrate usage,
colorant usage, parts replacement, etc. divided by average annual
volume to arrive at unit cost. CONSTpR has two different values for the

case of digital and conventional printing, $7 per day for conventional

51

and $30 per day for digital (Emden, 2001-2003). Therefore, the
printing setup cost will equal the cost constant for printing setup

multiplied by number of brands and number of global markets

combined.

CONSTPR (conventional) = ($7 x 365) / V (3.4a)
CONSTPR (digital) = ($30 x 365) / V (3.4b)
P5 = (Nb + Nm) X CONSTps (3.5)

CONSTps = Printing Setup Cost Constant

CONSTps represents the annual expenses on labor and materials
for setup for each production run (double sided tape production in the
case of conventional printing amounting to a representative value of
$800 and color setup mostly in the form of labor in digital amounting
to a representative value of $250). Therefore, printing run cost equals
printing setup costs added to volume multiplied by the printing run

cost constant, which is summarized in equation 3.6.

PR = ((Nb + N...) x CONSTPs) + (CONSTPR x V) (3.6)

52

Calculation for Inventory Cost

The inventory cost (INV) is the value of inventory, which is
driven by the inventory carrying/warehousing cost activity under the
printing run activity category. The product factor value of product-
package system (Vp) is a direct factor that determines the value of
inventory. Cost of inventory can be calculated by multiplying average
annual inventory volume by value of product-package system (unit
price of product) by the inventory carrying percentage cost constant
(equation 3.7). Average annual inventory volume is reached by adding
safety stock to order quantity divided by two, which is the averaging

factor (equation 3.8). Equations below illustrate these statements.

INV = INVavg X Vp X CONSTICC (3.7)
INVavg = Average annual inventory volume

SS = Safety Stock (see Calculation of Stock-out Costs section below)
CONSTICC = Inventory carrying cost percentage factor (constant)

CONSTICC = 8% (time value of money tied up in inventory annually)

INV= ((55 + (OQ/2)) x CONSTICC x vp (3.9)

53

Calculation for Stock-out Cost

Stock-outs happen when consumers cannot find the product on
the shelf at the point of purchase. This could result in potential lost
sales, if the consumer instead purchases a competitive product
instead. In addition to the potential lost sales, the stock-out costs
include intangible components such as loss of goodwill and potential
delays in other parts of the system (Nahmias, 2001, pg 264).

Stock-out cost is usually very hard to determine, and therefore
service level can be used as its replacement (Nahmias, 2001).
Quantifying service levels can be done in numerous ways, suggest
Graman (1999) based on Fogarty et al (1991). However, service level
generally refers to either the proportion of demand satisfied directly
from the shelf (i.e., fill rate) (Graman, 1999), or the probability of not
stocking out (Graman, 1999). Again, since most managers interpret
stock-out as the proportion of demand met directly from stock, it is
proper to use fill rate as the measure of service level (Nahmias, 2001).

Sox et al. (1997) addresses the issue of filling orders to meet
the targeted service levels in a fixed length of service window.
Therefore, it is necessary to identify variables representing the stock-
out period, and the service level objective (SL). Graman (1999)
expresses the proportion of demand satisfied directly from the shelf in

a given stock-out period as expected number of stock-outs divided by

54

the stock-out period. Therefore, the fraction of demand that is

satisfied from the shelf would be one minus that. When we isolate the

expected number of stock-outs and multiply by the profit margin (Mp)

for each unit we arrive at the cost of stock-out as expressed in

equation 3.10. However, due to the sensitivity of the profit margin

data, the case study subjects were asked to provide daily stock-out

costs instead (STOpd). Therefore, as illustrated in equation 3.11,

annual stock-out cost equals stock-out cost per day multiplied by the

average annual volume times one minus service level.

STO = CONSTSTO x (100 - SL) x Mp
STO = 5mm x v x (1-SL)

f(k) = (1-SL) x (OQ/Ddctstd)

55 = k x Ddct std

Ddct std = Standard deviation of demand during cycle time

 

Ddct std = JDDavg 2 x CTstd2 + CTavg x Dstal2

DDavg = Average daily demand = V/365
CTstd = Cycle time standard deviation = CTavg x 0.005
CTavg = Average cycle time = 365/Npr

Dstd = DD,vg standard deviation = DDavg x 0.005

55

(3.10)

(3.11)

. '1‘!

|"'." ‘ r.‘

Calculation for Obsolescence Cost

Obsolescence (OBS) plays a major role in increased incentives to
postpone labeling to the distribution center level. The product factors
that impact obsolescence are frequency of regulatory changes (Fm),
frequency of marketing based changes (ch), and frequency of
package engineering design changes (ch). Obsolescence is a cost
category driven by the discretionary activity of abnormal scrap. This
involves disposing of the scrap and rework. Rework cost is the same as
the printing cost explained previously (see equation 3.3). Therefore,
the cost of obsolescence is the cost of printing run multiplied by the
sum of the product factors that contribute to the obsolescence of
labels divided by the number of annual production runs. These are
regulatory changes, marketing based design Changes, and package
engineering design changes. These statements are illustrated in the

following equations.

OBS = (CONSTobs + PR) x ((Frc + ch + ch)/Npr)) (3.12a)
CONSTobS = Scrap Handling / Disposal Cost Constant

OBS = PR x (F.c + Fmc + ch)/Npr) (3.12b)

The case study interviews revealed that CONSTobs is insignificant

compared to the cost of rework. It also tends to be hidden, usually in

56

 

different departments within a company, which makes it harder to
quantify. Table 3.5 summarizes the total cost formulas and variables.
Next, the seven product factors are discussed including their effects on

cost factors and their identified ranges of values.

 

TABLE 3.5

 

Summary of Total Cost Formulas and Variables

 

 

Tc = PR+INV+STO+OBS

PR = PS + (CONSTPR X V) = ((Nb '1' Nm) X CONSTps) + (CONSTPR X V)
INV = ($5 + (OQ/2) x Vp x CONSTICC

STo = STopd x V x (1-SL)

035 = PR X (Fm + ch + ch)/Npr)

TC = Total cost of printing labels
PR = Printing run cost

P5 = Printing setup cost

INV = Inventory cost

STO = Stock-out cost

OBS = Obsolescence cost

Frc = Frequency of regulatory changes

Frc = Frequency of marketing based design Changes

Frc = Frequency of package engineering design changes
= Value of product-package system

Nb = Number of brands

Number of global markets

Np, = Number of annual production runs

0Q = Order quantity = V/ (Npr /2)
V = Average annual volume
CONSTpR = Printing run cost constant
CONSTpR (conventional) = ($7 x 365) / V
CONSTPR (digital) = ($30 x 365) / V
CONSTPS = Printing setup cost constant
CONSTICC = Inventory carrying cost constant (percentage)
= 8% (time value of money tied up in inventory annually)
SL = Service level
STOpd = Stock-out cost per day
f(k) = (1-SL) x (OQ/Ddctstd)
$5 = k X DdCt Std
Ddct std = Standard deviation of demand during cycle time

Ddct std = ‘/ DDavg 2 X CTstd 2 + C Tavg x Dstd 2

DDavg = Average daily demand = V/365

CTstd = Cycle time standard deviation = CTavg x 0.005
CTavg = Average cycle time = 365/Npr

Dstd = DDavg standard deviation = DD,”g x 0.005

2
3
ll

 

 

57

 

PRODUCT FACTORS

In this section, each product factor is defined and the effects of
each factor on the total cost of printing labels are discussed. The
ranges of values for the cost and product factors are presented in the
last section of this chapter titled Methodology and Research Design.
The data gathered through the case study interviews are used to
establish the representative minimum, mode, and maximum ranges of
values for the product factors and some of the cost factors as indicated
previously. Therefore, the data set is not intended to represent any
one specific case.

Each of the seven product factors has a distinct effect on the
total cost of printing labels, which is presented in the section titled
Model Structure. Depending on a given business environment as
characterized by the collective effect of these product factors, the total
cost of printing labels may increase or decrease. Similarly, as the
value of the cost factors that compose the total cost of printing labels
are varied, the incentive to postpone package labeling to the
distribution center level will be higher or lower. Distribution center
level represents the last leg of the supply chain where large-scale
quality printing can occur. Again, the effects of various business

situations (caused by these product factors) on total cost and the

58

 

decision to postpone package labeling to the distribution center level
are captured by these seven product factors.

As previously discussed, this study approaches the problem from
the total cost perspective, since the goal for any business should be to
reduce total cost. As explained in Chapter II, all these variables are
determinant factors for implementing postponement strategies from
the perspective of minimizing total cost. Postponement strategies also
bring in flexibility to the production processes, as explained in previous
chapters. However, in this study the focus is on minimizing total cost.
Since postponement has an effect on total cost, the impact of each of
the factors is investigated to identify their relative effects on cost
components that make up the total cost of printing labels. Next, each
product factor is defined and its effects on the cost of printing labels

are discussed.

Frequency of regulatory changes (Fm)

This factor refers directly to the regulations and hence regulatory
Changes imposed by the Food and Drug Administration (FDA). As a
result of numerous exploratory interviews with The John Henry
Company, this emerges as one of the major challenges faced by the
pharmaceutical industry especially in launching of

compounds/products. Due to the sensitivity of the timing of the

59

 

launch, companies sometimes take the risk of printing labels by
speculating on the FDA regulations. The resulting effect of this factor
on total cost of label printing and the decision to postpone is
important. This factor results in obsolescence and an increase in costs
for the printing setup activity due to the need to rework the
setup/programming to comply with the changes each time they occur.
Hence, as the frequency of changes increases, the cost of setup should
increase. In addition, costs for printing run activities should also
increase since most upgrades involve more information to be provided
on a label (Emden, 2001-2003). This results from the costs incurred
for the existing inventory of outdated labels. In addition, there should
be an increase in abnormal scrap/rework activity related costs, but no
changes will be observed for the special requirements cost category.
As determined through case study interviews, the range of

representative values for this factor is one to three, averaging at two.

60

 

Frequency of marketing based design changes (Fm)

Sometimes, label design is modified in the middle of a product
launch, or heavy sales, driven purely by marketing factors. This
condition results in increased numbers of abnormal scrap driving the
total cost up. The effect of such situations on total cost and the
decision to postpone is captured by this factor. This will have a similar
effect on the cost drivers as the above parameter, since labels will
need to be redone. Therefore, an increase in obsolescence will be
observed. In addition, the colorant usage rate should increase due to
the trend of employing more colorful and bright designs and more
expensive ink (such as metallic ink), which permeate into updated
label designs (Emden, 2001-2003).

Case study interviews revealed the range of values for this
variable to be one to five, averaging at three. The reason for values to
be less when compared to frequency of regulatory changes is mostly
due to the relative flexibility of the change. Usually, when marketing
based change occurs for a label, the cost of obsolescence is provided
to the marketing department for the cost benefit analysis of making
the change. Based on this, the Changes may be implemented between
production runs, which does not impact the total cost of printing

labels.

61

 

Frequency of packaging engineering design changes (ch)

This factor is similar to ch, in that label production may need to
be modified in order to accommodate changes in the packaging of the
product. An example for this is the introduction of flexible pouch
packaging for tuna fish. The information on the label is the same,
however, due to the change in the package, label production needs to
be altered. Therefore, the resulting effect of this factor on the total
cost of label printing and the decision to postpone is important to
explore. When the industry experiences higher probability package
engineering design changes, the impact on the cost drivers should be
similar to higher rates of regulatory changes. Most package
engineering design changes involve more complicated designs such as
more sophisticated folding to better prevent shock and Vibration
damage (Emden, 2001-2003). This inherently increases the cost of
printing. There is also the probability of folding machines to be
jammed more often, increasing the abnormal scrap/rework costs. The
argument for the colorant usage rate may not hold since package
engineering design change does not necessarily translate into more
colorful labels. Obsolescence remains to be a factor that increases
with higher frequencies of package engineering changes.

As determined through case study interviews, the representative

range of values for this variable is zero to two, averaging at one.

62

 

Similar to the frequency of marketing based changes, the decision to
implement the change can sometimes be postponed so as not to affect
the total cost. In addition, package engineering design change is a
value-driven change. Therefore, most of the time the goal is to
minimize cost, which provides incentive for the firm to synchronize
such Changes with the other changes (i.e., regulatory changes or
marketing based changes).

Table 3.6a summarizes the statements above by illustrating the
product factor frequencies of regulatory, marketing based, and
package engineering changes, their associated value ranges minimum,
mode, and maximum, and their relation to cost factors, namely
printing run, inventory, stock-out, and obsolescence.

TABLE 3.6a
Product Factors Associated Ran es and Their Relation to Cost Factors
F F F
Factors

NV

BS

ated Ran es
inimum
ode
aximum

 

Value of product-package system (Vp)
The value of the product has been documented to be a factor in

postponement (Zinn, 1986). This thesis expands the investigation

63

 

further to include the product-package system. Consequently, the
effect of the value of product-package system on total cost and the
decision to postpone is captured by this factor.

Zinn (1986) discovered that postponement of the labeling
function to the warehouse level is a proper solution for products with
higher value. These findings indicate that the higher the value of a
product the higher the incentive to postpone labeling to the warehouse
level (Zinn, 1986). The assumption is that the cost of carrying
inventory of products with high values is also high; therefore,
postponement remains to be a valid strategy. In this study, the
variable is expanded to encompass the value of product-package
system. The Zinn model of thinking should still be valid. In addition,
the case should further be explored and measured through case study
interviews if high value product-package systems require more
attractive labels with complicated designs and more color. Therefore,
an increase in cost for colorant usage rate is also expected in addition
to the cost of carrying inventory. Printing setup should not vary, since
the value of the product-package does not necessarily imply multiple
setups.

The case study interviews revealed representative ranges of
values for Vp to be $0.5 on the low end and $2 on the high end,

averaging at $1.

64

 

Table 3.6b illustrates the product factor, value of product-
package system, its associated value ranges, and its relationship to

the cost factors.

TABLE 3.6b
Product Factors and Associated Ran es
F F F V
Factors

NV

BS

ated Ran es
inimum
ode
aximum

 

Number of brands (Nb)

Number of brands is an important factor especially when the
scale is large. Having a large number of brands translates into having
to do multiple setups for label printing jobs, which in turn has an effect
on total cost of label printing. Depending on the number of brands, a
company may need to print different kinds of labels (i.e., different
print-job setups) for the same compound, since they are marketed
differently. Therefore, the resulting effect of this factor on the total
cost of label printing and the decision to postpone is incorporated into
the cost model.

Number of brands also includes different indications of a

compound, since the labels will be different for each indication. Large

65

 

numbers of brands translate directly into increased activity in printing
setups due to different label requirements for different brands, even
though the brands may use the same base formula (compound). In
addition, the more brands the shorter the length of each labeling run
which results in higher cost of postponement (Zinn, 1986). On the
other hand, with more brands, the opportunity to consolidate safety
stocks becomes lower, which increases the cost of speculation relative
to the cost of postponement (Zinn, 1986). Zinn (1986) concluded that
the latter criteria turned out to be stronger therefore there was a
greater incentive to postpone labeling.

From the case study interviews, it is evident that the number of
brands a pharmaceutical company carries on a particular compound
can be as little as five or as many as 20, averaging at 12.

Table 3.6c summarizes the statements above by tabulating the
product factor number of brands, its associated value ranges, and its

relationship to the cost factors.

TABLE 3.6c
Product Factors Associated Ran es and Their Relation to Cost Factors
F F F v N

Ra
inimum
ode
aximum

 

66

 

Number of global markets (N...)

An apparent effect of this factor is the need to deal with multiple
languages. Having multiple languages translates into having to do
multiple label printing setups for each language. Similar to the number
of brands, this factor affects total cost of printing labels and the
decision to postpone.

The same effect from greater number of brands should be
expected from larger number of global markets primarily due to the
increased number of languages the process has to accommodate. As a
summary, printing costs are expected to increase with more global
markets.

The case study interviews indicate that the number of global
markets that a pharmaceutical company generally operates in can be
one or four, averaging at three. Table 3.6d illustrates the product
factor number of global markets, its associated value ranges, and its
relationship to the cost factors.

TABLE 3.6d
Product Factors Associated Ran es and Their Relation to Cost Factors
F F F V N N
Fa rs

NV

BS

immum
mm
ammum

67

 

Number of annual production runs (N,,)

This factor is mostly driven by the seasonality of product (Sp). In
the case of a stock-out of a product with seasonal demand (such as
allergy medication), company may lose consumers for a longer period
of time, depending on the length of the season. This, in turn affects
the decision to postpone and the total cost of printing labels.

As the number of annual production runs increase, the cost of
holding inventory decreases. This is primarily due to the ability of
satisfying the demand by less inventory. Assuming a constant annual
demand, a company may choose to produce the demand in one run
versus weekly runs. In the first case, the annual average inventory will
be demand divided by two, assuming linear expenditure of inventory.
In the latter case, the company produces demand divided by 52
(weekly demand), and therefore the annual average inventory will be
demand divided by 52 divided by two. Similarly, as the annual
production runs increase, the abnormal scrap/rework costs will
decrease due to less inventory being subject to obsolescence.
However, the cost of stock-outs increase due to reduced inventory
holdings.

The case study interviews indicate that production can happen as

often as weekly or in very rare cases (seasonal) two times a year. This

68

assigns the lowest value for this variable to be two and the highest 50,
averaging at 12.

Table 3.6e tabulates the product factor number of annual
production runs, its associated value ranges, and its relationship to the
cost factors.

TABLE 3.6e
Product Factors Associated Ran es and Their Relation to Cost Factors
F F F V N N N
Fa

NV

BS

ciat Ran es
inimum
ode
aximum

 

MODEL STRUCTURE

The statements summarizing how the seven product factors
relate to the total cost of printing labels are illustrated in Tables 3.7
and 3.8 below. The direction of the arrow indicates the relationship
between product factors and cost categories (i.e., the up-arrow
specifies a directly proportional relationship and the down-arrow
indicates an inversely proportional relationship between the product

factor and the corresponding cost category). In addition, the seven

69

product factors relate to the ABC model of cost activities as illustrated

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

in Table 3.8.
TABLE 3.7
Effect of Product Factors on Cost Factors of Printing Labels
PR INV STO OBS
Fr; 1‘ t
ch t t
ch 4* t
Vp m 4‘
Nb 1‘
Nm 1‘
N... t 4‘
TABLE 3.8
Cost Activities versus Product Factors
Frc ch Foc VD Nb Nm NQL
1. Direct Activities
A. Printing Setup + + + 0 + + +
B. Printing Run + + + + 0 0 O
1. Substrate Usage Rate 0 0 0 + 0 0 0
2. Colorant Usage Rate + + + + 0 0 0
3. Equipment Parts Replacement Rate 0 0 0 0 0 0 0
4. Inventory Carrying/Warehousing 0 0 0 + 0 0 -
2. Discretionary Activities
A. Abnormal Scrap/Rework i + [+ i + I 0 I 0 i 0 i 0 J
3. Special Requirements Activities
A. Special Customer Requests i O i 0 i 0 i 0 i 0 i 0 I 0 i

 

 

([0] denotes no change, [-1 denotes a decrease in the cost activities as the
product factors increase, and [+] denotes an increase in the cost activities as the
product factors increase.)

These illustrations summarize the product factors and their
effects on the total cost of printing labels. The next section discusses

the research hypotheses.

70

RESEARCH HYPOTHESES

New technologies, such as electrophotographic/digital printing,
bring greater incentives to implement postponement strategies in
package labeling. The purpose of this research is to develop an
economic model for implementing postponement strategies. The study
presents a model that includes seven product factors and uses a total
cost approach in analyzing the effects of these factors on the total cost
for printing labels and develops economic justification for
postponement in various cases.

The question that the hypotheses address is: How does each of
the seven product factors impact the total cost for printing labels and
the decision to postpone? Below are the expected effects of each of
the seven product factors has on the decision to postpone, namely

propositions for this study.

A. The HIGHER the frequency of regulatory changes (Fm), the
GREATER the incentive to postpone labeling to the distribution
center level, and the GREATER the cost of obsolescence and the
cost of printing runs (most upgrades involve more information to be

put on labels).

71

. The HIGHER the frequency of marketing based changes (ch), the
GREATER the incentive to postpone labeling to the distribution
center level, the GREATER the cost of obsolescence and the cost of
printing runs (most changes involve more color to be put on labels;
or more expensive colorant, such as metallic ink).

. The HIGHER the frequency of packaging engineering design
Changes (Ppc), the GREATER the incentive to postpone labeling to
the distribution center level, the GREATER the cost of obsolescence
and the cost of printing runs (most changes involve more
complicated designs such as folding to better prevent shock and
vibration damage).

.The GREATER the value of product-package system (Vp), the
GREATER the incentive to postpone labeling to the distribution
center level (generally, as the value of product-package system
increases the cost of printing labels increases).

.The GREATER the number of brands (Nb), the GREATER the
incentive to postpone labeling to the distribution center level.

. The GREATER the number of global markets (Nm), the GREATER the
incentive to postpone labeling to the distribution center level.

.The GREATER the number of annual production runs (Npr), the

LOWER the incentive to postpone labeling to the distribution center

72

level (the GREATER the cost of stock-outs but the LOWER the cost

of inventory).

Distribution Center Level as referred here, represents the last leg
of the supply chain where large-scale quality printing can occur. Figure
3.1 presents a diagram of the research hypothesis explained above.
These hypotheses are tested using the methodology explained in the

next section.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FIGURE 3.1
Research Hypotheses Diagram
Frc
-:$ PR
ch
+
INV
ch ___> ‘5
Postponement
Effectiveness
VP J
y STO
Nb
P OBS
Nm
NPr ,_

 

 

 

METHODOLOGY AND RESEARCH DESIGN

This section explains the case study interview method used to
develop the model and verify ranges for the product and cost factors.

Next, the cost simulation is discussed in detail.

CASE STU DY INTERVIEW

Case study research method is utilized to explain each product
factor and their relationship to the total cost of printing labels. This
way, the context of label printing is more comprehensively
investigated. For example, Yin (1994) indicates that survey, as an
alternative research strategy, has extremely limited ability to
investigate context.

The case study is conducted in the form of interviews. The case
study subjects who are invited to participate in the study are mostly
alumni or other contacts from Michigan State University’s School of
Packaging and contacts and customers of The John Henry Company. In
order to effectively assess the relationship between the product factors
and total cost of printing labels, a particular industry is chosen. Due to
the seemingly amplified effects of the product factors, the primary

business selected is the pharmaceutical industry.

74

A set of exploratory interviews is performed following formal
interviews. First, the interviewees are asked verify the product factors
and their effects on the total cost of printing labels. Then, as explained
previously in this chapter (Model and Theory Development: Product
Factors section), the case study participants provide three ranges of
values for each product factor (i.e., minimum, mode, and maximum).
The data gathered through the case study interviews are then used to
establish the representative ranges of values for the product factors
and average annual volume. Table 3.9 illustrates these values.

TABLE 3.9
Product Factors and Associated Ranges

F F F V N N N V
inimum 0. 50000

Ode 1. 1 20 000 00
aximum 2. 2 100 000

 

The cost model for printing labels is constructed starting from
label quantities of 1,000 up to 10,000 linear feet in 1,000 increments.
The reason for this stems from the fact that the economical run length
for digital printing for labels seems to be within a range of 2,000
(Emden, 2001-2003) to 7,500 linear feet (Bence, 2001). Somewhere
within this range, the cost function of litho intersects with the cost
function of electrophotographic/digital and takes a downward trend
while the cost function for digital moves upward with respect to total

cost. This statement is illustrated in Figure 3.2 with hypothetical

75

numbers. Even though Zinn (1986, p 178) indicates that demand
levels do not have any effect on the decision to postpone, this
research still takes the demand levels into account when analyzing
total cost as articulated above. It should also be noted that the study
takes into account the varying ways of measuring label production
quantity. The preliminary interviews suggest that linear-feet is the
favorable measure, however the interviewees are given the
opportunity to comment on this.

FIGURE 3.2
Cost Functions of Lithographic versus Digital Printing

 

+Dlgltal i
+Litho j

 

 

 

 

Total Cost of Label Printing

1'

2,000 linear feet according to Emden and 7,500 linear feet according to Bence.

 

 

The case study subjects are then asked to build and explain a
business case for each possible business scenario. During this process,
the effect of the production volume is also considered to explain which
of these factors has the most impact on total cost with each level of

production volume.

76

The data collected through the case study interviews are used to
establish the representative ranges of values for the cost factors, the
variables that are used to calculate them, and the product factors. The
representative ranges of values are then used to validate the research
hypothesis, to develop a total cost model for printing labels, and to
quantify the effects of postponement. This consists of validating the
data using a spreadsheet software application using Microsoft® Excel®.
The section titled Cost Simulation explains the method in detail.

When litho printing methods are used, setup costs are higher
with respect to digital printing and variable costs decrease with the
increased quantity of labels produced (Emden, 2001-2003). In the
case of digital printing however, setup costs are observed to be less
than litho printing setup costs, and variable costs stay constant
regardless of the quantities of labels produced (Emden, 2001-2003).
These statements are summarized with representative numbers in

Figure 3.2 (Emden, 2001-2003) above.
COST SIMULATION

In conducting the investigation, cost model developed earlier for
package-labeling system is built into the simulation. The data collected

through the case study interviews are used to establish the

77

representative ranges of values for the cost factors, the variables that
are used to calculate them, and the product factors. A combination
simulation-regression analysis is conducted using the data obtained
through the case study interviews. The simulation model is developed
to observe the impact of all the independent variables on the cost
drivers and the package labeling system cost for digital printing and
for conventional printing. A multiple regression analysis is then
executed in order to identify the independent variables among the cost
drivers so as to isolate the ones with the greatest impact on
postponement. A package labeling supply chain system profile is
developed to reveal when savings can be achieved. The effects of
postponement on each of the seven product factors are explored with
a focus on minimizing total cost utilizing the spreadsheet model.

Next, the effects of production quantities on the ABC cost factors
are investigated. As explained previously, the support activities are not
taken into account since a facility must run whether it produces 10,000
or 1 million linear feet of labels at any given time frame. Therefore,
the category of operating costs is the group under consideration.

Finally, the individual and relative impact of the seven
independent variables listed as frequency of regulatory Changes (Fm);
frequency of marketing based changes (ch); frequency of package

engineering design changes (ch); value of product-package system

78

(Vp); number of brands (Nb); number of global markets (Nm); and
number of annual production runs (Npr) on total cost should be
analyzed. The fact remains that the operating activity cost items is the
contributing cost activity in this analysis.

As explained in the section titled Model and Theory Development
and summarized in Table 3.8, the items listed under printing run under
direct activities as Substrate Usage Rate, Colorant Usage Rate, and
Equipment Parts Replacement Rate, are not expected to be affected
with increased/decreased levels of any one of the seven independent
variables, namely product factors. This argument also holds for the
Special Requirements Activities. Further, it is observed that an
increase in either number of brands or number of global markets
results only in significantly increased levels of the cost item listed
under direct activities as printing setup. On the other hand, increased
levels of Printing Setup, Inventory Carrying/Warehousing, and
Abnormal Scrap/Rework should be recorded as the frequencies of
regulatory (Frc), marketing based (ch), and package engineering
based design changes (ch) increases.

The goal is to identify the factor(s) with the most impact on
postponement, as it relates to total cost. This way, the area where
postponement has the most impact on total cost is isolated. To

achieve this, regression analysis is run on the dependent variable and

79

contains the independent cost factors. The dependent variable is the
difference of total cost of printing labels between postponement
(digital printing) and speculation (conventional printing) strategies.
Therefore, a negative difference favors using digital printing method.
Digital printing is acknowledged as an enabler to postponement in
Chapter 11 under the section titled Analysis of Label Printing. Hence, a
negative difference suggests implementing postponement as package
label printing strategy.

The cost simulation is run for two different sets of models; one
representing the cost model of printing labels using digital technology
and the other representing the cost model of printing labels using
conventional printing method. This is to capture the possible conditions
under which postponement strategies are justified. The collection of
settings under which postponement strategies are the least expensive
can be expressed as a relevant range for using digital printing
technology. As explained previously in this chapter under the section
titled Product Factors, the minimum, mode, and maximum data values

are collected using case study interviews, as illustrated in Table 3.10.

 

 

 

 

 

TABLE 3.10
Examples of Variables and Associated Ranges
Variable 1 Variable 2 Variable 3
Minimum 1 4 7
Mode 2 5 8
Maximum 3 6 9

 

 

 

 

 

80

A triangular generating function in Microsoft® Excel® is used to
generate individual observations for each of the seven variables (i.e.,
product factors). With a normal distribution, the data alleges infinity,
which is not the case in this study (Closs, 2003). In addition, the
triangular function provides us with the opportunity to evaluate non-
symmetric distributions, which allows the possibility to have the values
range freely without forcing symmetry.

The triangular generating function assumes a cumulative
distribution for a triangular probability density function and takes
random values of each variable. In order to ensure statistical
significance, one thousand observations are generated using this
function. Each observation represents a unique combination of the
seven product factors with each factor independently following a
triangular probability distribution. Appendix B lists the first twenty-five
observations in the simulation model. As pointed out previously, for
each observation, the cost of printing digital and conventional are
calculated and the difference of the costs indicates the favorable
method. Table 3.11 illustrates this method with ten observations as
sample. In this illustration, variable 1 has minimum 1, mode 2, and

maximum value of 3.

81

TABLE 3.11

Exam le of Cost Simulation

Variable Variable Variable Variable Tc Tc Tc (digital-conv.)
1 min 1 mode 1 max 1 di ital conv.

 

Even though the seven product factors are treated as
independent variables, it must also be noted that many other factors
are likely to affect the total cost of printing labels, such as average
annual volume (V). Therefore, the model also incorporates average
annual volume as the eighth variable and treats it as an independent
variable. In addition, in order to account for the additional factors that
may be influential as probable independent factors, an error term is
used. Therefore, the error percentage term introduced previously in
this chapter addresses the additional factors that may not have been

taken into account.

CONCLUSION

In this chapter, the groundwork for developing an economic
model for package label printing in pharmaceutical industry is
described. The next Chapter presents the data analysis according to

the methodology explained above.

82

CHAPTER IV

DATA ANALYSIS

A total of seven case study interviews are conducted for
identifying the representative minimum, mode, and maximum values
for each product factor and all cost variables that compose the cost
factors which, when added, make up the total cost (see equation 3.1).
Appendix A lists the interview questions and discussion points that
were addressed by the case study participants. Data on some of the
cost variables was not available due to confidentiality or difficulty in
compiling the data especially when multiple departments were
involved. However, the case study interviews are used to identify the
representative ranges of the data set, which is not intended to
symbolize any one specific case.

The simulation model is run based on the data as gathered and
utilized to set the representative ranges of the variables. After the
simulation of the model, sensitivity analysis is performed to measure
the scale of significance for each product factor.

This chapter presents the data analyses as explained in the cost
simulation section of Chapter III and the results for each product

factor and the multiple variable data analysis to isolate the product

83

factors with the significant impact on total cost for printing package
labels in the pharmaceutical industry.

As a summary, the data collected through the case study
interviews are used to establish the representative ranges of values for
the cost factors, the variables that are used to calculate them, and the
product factors. A combination simulation-regression analysis is
conducted using the data obtained through the case study interviews.
The simulation model is developed to observe the impact of all the
independent variables on the package labeling system cost difference
between digital and conventional printing. A negative result indicates
that digital (plateless) printing provides cost reduction with respect to
conventional (plate) printing. Observations on over-the-counter (OTC)
drug, prescription drug, and surgical equipment products are captured
during the interview process and are used to create different scenarios
for the sensitivity analysis.

Multiple regression analysis is executed in order to identify the
independent variables with statistical significance on the dependent
variable, namely the difference of total cost of printing labels digitally
versus conventionally (Tc). The independent variables analyzed include
the seven product factors (i.e., Frc, ch, ch, Vp, Nb, Nm, and Npr) and
the average annual volume (V). A package labeling supply chain

system profile is developed to reveal when savings can be achieved.

84

The effects of postponement on each of the seven product factors are
explored with a focus on minimizing total cost utilizing the spreadsheet
model.

Finally, the relative importance of the product factors is
examined. As a result, the product factors are ranked in descending
order of impact on the difference of total cost of printing labels

digitally versus conventionally.

EFFECT OF FREQUENCY OF REGULATORY CHANGES

Figure 4.1 illustrates the one thousand observations of the
independent variable frequency of regulatory changes (Fm) with
respect to the dependent variable the difference of total cost of
printing labels using digital versus conventional printing technology
(Tc). The multiple variable regression analysis illustrated in Table 4.1
indicates that Frc with a p-value of 0.579069861 (higher than 0.05) is
NOT a significant factor affecting the total cost of printing labels.

This result could be due to the fact that the ranges of values for
FIrc are not high compared to ch, which is statistically significant, as
explained in the next section. This is tested with a sensitivity analysis,
where the values of the product factors with high p—values are
increased, for example, doubled. When this analysis is performed, Frc

becomes a significant variable affecting the total cost of printing

85

labels. The resulting multiple variable regression analysis is presented
in Table 4.2a. The simulation and multiple variable regression analysis
confirm significance when the ranges of values for Frc are otherwise
increased, such as making the values equal to the ranges of values for
ch.

TABLE 4.1

Multiple Variable Regression Analysis

 

Regression Statistics

 

 

 

 

R Square .727
Coefficients t Stat P-va/ue
Intercept 229422.891 11.422 .000
Frc -2625.589 -.555 .579
ch -7598.251 -3.262 .001
ch -2753.274 -.566 .571
Vp -2232.136 -.361 .718
Nb -14771.437 -22.405 .000
Nm -15943.307 -5.071 .000
Npr -8762.946 -46.370 .000
V .000 .991 .322
TABLE 4.2a

Multiple Variable Regression Analysis: FrC ranges of values doubled

 

Regression Statistics

 

 

 

R Square .740
Coefficients t Stat P-va/ue

Intercept 273545.524 14.181 .000
Fm -6415.972 -2.613 .009
FmC -4838.225 -1.959 .050
ch -8882.011 -1.836 .067
Vp -19886.992 -3.190 .002
Nb -16871.899 -26.446 .000
Nm ~14801.946 -4.867 .000
Npr -8622.966 -45.747 .000
V .000 -. 127 .899

 

86

ABC 329.com» mczcto .mcozcoEoo msmco> .336 05m: m_onm_ 9555
Go umou BUB Co DEGREE DE 8 uuoamm: 5:5 AB“; momcmco 383309. 00 55:08.. 00 mcozmtmmno 0004

 

08.85.
88.88-
88.83.
08.8%-
. 88.88-
J 80.8%-
c. 58.85.
L a

1., 08.8 5
on mm on m; . Sodom»

mm 000.0000.

(Iauonquuoo-MIBIO)
Isoo mo; )0 aousIaula

 

9.555 .055 ammo—own. “—0 «moo .30... co 0.."— po nong—

 

 

 

mEuELn. .003 ommxumu no #30 .59.. co mom—5:0 Baum—:03. no 5:03—09."— uo HUNGEH

H.¢ mmbwnu

87

EFFECT OF FREQUENCY OF MARKETING CHANGES

Figure 4.2 illustrates the one thousand observations of the
independent variable frequency of marketing changes (ch) with
respect to the dependent variable the difference of total cost of
printing labels using digital versus conventional printing technology
(Tc). The multiple variable regression analysis illustrated in Table 4.1
indicates that ch with a p-value of 0.001146057 (lower than 0.05) IS
a significant factor affecting the total cost of printing labels.

As displayed in Table 4.2a, doubling the variable ranges for Fm
renders ch as insignificant only marginally, (p-value of 0.050348782).
However, when the variable ranges for Pm and Fm are made equal (i.e.
1, 3, 5), then both variables become significant. The resulting
regression analysis is displayed in Table 4.2b. This indicates that under
certain value ranges, the variables become insignificant factors

affecting the total cost of printing labels.

88

TABLE 4.2b
Multiple Variable Regression Analysis:

Frc ranges of values increased to equal to ch ranges of values

 

Regression Statistics

 

 

 

R Square .730
Coefficients t Stat P-value

Intercept 215371.178 11.785 .000
Frc -10938.540 -4.518 .000
ch -514S.315 -2.192 .029
ch -3626.316 -.782 .434
Vp 1545.241 .244 .808
Nb -14658.612 -23.235 .000
Nm -10593.274 -3.438 .001
Npr -8255.826 -45.667 .000
V .000 -1.0197 .308

 

89

.35 320568 mcscta _mco_uc0>coo mszm> 5506 05m: m_0n_m_ 0:553

Co 58 5.8 .0 06:20.20 05 B 8030.. 5:5 Ace": mmmcmco 0:59:92 .6 5:03.800 .0 mcosmtmmno 000;

 

 

   

 

 

 

2.:
80.88- a
.- . o . s l .. l. 80.85. m m
o 1 w. o 9 . . - .. o o . 88.88- .M m
.. .3 i $19.93. 9/9 o. o... o. s .o.ooo.8m».ml.m
6 ’9 O . .. .Jl Oi i . . .. . 00" ROWI {I l i 00°50th Wu 9
E ”9.4M 9 ._ . . .. . .. «iofi .bl 80.88- u w
9 l A I.
. 999i9 9fiwg \& 999 #93 909.9 .9. 0. WtoooWooww W O
omr 93992.0... n58°85. am
9 9.89 o 99 . . , _ O
- . it- . o o 8 w m
i . - . ..... o . i 4 8° 85 (I. 3
Q... m... 3. 3 em 3 od 3 88.8%
mass... .33 .o .30 .83 :o 9E .6 SEE.

 

 

 

0:35.... .053 009.03. no 500 .30... co m0mcmsu 0:300:22 00 5:03.92". so UNQEH

N6 man-0H".

90

EFFECT OF FREQUENCY OF PACKAGE ENGINEERING CHANGES

Figure 4.3 illustrates the one thousand observations of the
independent variable frequency of package engineering design
Changes (fix) with respect to the dependent variable the difference of
total cost of printing labels using digital versus conventional printing
technology (Tc). The multiple variable regression analysis illustrated in
Table 4.1 indicates that ch with a p-value of 0.571438514 (higher
than 0.05) is NOT a significant factor affecting the total cost of printing
labels.

Similar to the argument made previously for Frc, this result could
be due to the fact that the ranges of values for ch are not high
compared to ch, which is statistically significant. Once again, the
simulation and multiple variable regression analysis confirm
significance when the ranges of values for ch are increased, such as
making the values equal to the ranges of values for ch. When this
analysis is performed, ch becomes a significant variable affecting the
total cost of printing labels. The resulting multiple variable regression

analysis is presented in Table 42¢.

91

TABLE 4.2c
Multiple Variable Regression Analysis:

ch ranges of values increased to equal to ch ranges of values

 

Regression Statistics

 

 

 

R Square .739
Coefficients t Stat P-value

Intercept 268768.302 13.379 .000
Frc -7296.102 -1.599 .110
Pm -12543.999 -5.239 .000
ch -8768.439 -3.710 .000
Vp -2684.772 -.437 .662
Nb -15257.203 -23.S48 .000
Nm -16418.177 -5.165 .000
Npr -8633.146 -46.764 .000
V .000 -.538 .591

 

92

.1 .5 30.05.63 0:35.... _mco_uc0>coo m...m.0> $29.6 05m: 282 0:35;. u.o Dmou :33 .6 09.0th
05 8 6030.. 5.2. .8“: 8056 c930 05.50505 ommxomg .6 5:03.800 .6 mcosmtwmno oooq

 

 

 

 

 

 

(leuonuanuoo-Ieufila)
Isoo |BIO_|_ IO aouaIama

 

I. 8 85
on 3 m; 3 ~— 3 .8 o... no No 8088

8.2.5 .33 .6 $8 .28. co 8”. Co 83:.

 

 

 

0:35.“. .23.. no umou .30... :o 00052.0 anion. m:...00:.m:w 005.00g no >uc0aw0cu. «0 “SEE“

m..v NKDUHH.

93

EFFECT OF VALUE OF PRODUCT-PACKAGE SYSTEM

Figure 4.4a illustrates the one thousand observations of the
independent variable value of product-package system (Vp) with
respect to the dependent variable the difference of total cost of
printing labels using digital versus conventional printing technology
(Tc). The multiple variable regression analysis illustrated in Table 4.1
indicates that Vp with a p-value of 0.71790182 (higher than 0.05) is
NOT a significant factor affecting the total cost of printing labels.

Similar to the arguments made for the other product factors that
turned out to be insignificant, this result is likely due to the fact that
the ranges of values for Vp are not high enough to render it significant.
When tested with a sensitivity analysis by increasing the ranges of
values, Vp is rendered significant. This succeeded in depicting Vp as
significant only when the ranges of values were increased by about a
couple hundred folds (i.e., $5, $500, and $10,000 as minimum, mode,
and maximum respectively). The scenario under which these types of
values could be justified is in the case of surgical implants, or highly
complex pharmaceuticals such as cancer treatment drugs, because the
Vp values for such products tend to be much higher than more
common pharmaceutical products. The resulting multiple variable

regression analysis is presented in Table 4.2d.

94

TABLE 4.2d
Multiple Variable Regression Analysis:

Vp ranges of values increased to S, 500, and 10,000

 

Regression Statistics

 

 

 

R Square .018
Coefficients t Stat P-value

Intercept 3671293.261 .630 .529
FrC -1499685.226 -.981 .327
Pm -397873.461 -.520 .604
ch 912345.208 .589 .556
Vp 649.818 2.433 .015
Nb 152168.275 .733 .464
Nm -1745911.809 -1.740 .082
Npr -76777.932 -1.284 .199
V .071 2.425 .016

 

95

Aug-C 329.com: 9.05.... 58552.00 mzm$> 233.0 05m... 282 0:055
00 “moo :38 00 8:6th 9: 8 88mm: 5.3 13 E3m>m mmmxomabozuoa .6 o2m> no mcozmtwmno 000;

 

a>

4 8°88-
l l 80.85-
.888-
8°83-
8°81"-
888»-

 

 

 

 

 

(Inuonuanuoo-Ianfilo)
130:) mo; :0 aoualema

 

 

. 000.com»
0. r

0d Q— 0._. h; 0... m; ¢.— 0.? N._. F.—

9555 .33 .8 $8 .28 co n> .6 sane.

 

 

 

0:35....— _oaa._ no «moo .33- :o Eoum>m ommxuanbuavok. uo o:_m> ho yuan—EH

at? Han—0H".

96

.05 320533 0:355 _m:o3:o>:oo mame> $3.06 05m: m_onm_ 0:35.:
.o 88 538 .0 8:96.53 93 8 308mm. 53> 0.3 539$ 00963-5330.: .0 m_.._m> .o m:o3m>..omn.o coo;

 

   

 
  

 

. n>
7 . 0868.85.
0 . . ) a
.‘l: l - l‘i . 1 l! --- . 8o 08 88. a m
. . m a
7 . o . , o8 8o 88- u a
W u
83.. 08a 80.8 086. 308;.- 83. 80..” 83 83 0 80.08.85- 9 m
_ > - 7 C .9 H Q “0,". .M.‘ m. V§u . it... a.“ in. «111$ 0% m U
. . o o . 7 888.8; m m.
0 o . . N W
;l - o o -- , o8 O8 8% w o
O
7 . . l . 80.08.88 m m
08.9 83-. 83 o8.“ 80.8 08;.- 83. 08...” 83 08.. .. 2858.8g ..
7 0:35..."— Ena... no «moo .80... :o a> .o .035.
. l . - ll. . . ‘ lil- i- -

 

8863 .89... .3 mass. 2:9, ._>.
0:35...“— .on.~.. 30 $00 .mHOh :0 Ewum>m 00mxumn.-uu:_uo..n. no 0:.m> no «umaEH

QT? wKDUHu.

97

EFFECT OF NUMBER OF BRANDS

Figure 4.5 illustrates the one thousand observations of the
independent variable number of brands (Nb) with respect to the
dependent variable the difference of total cost of printing labels using
digital versus conventional printing technology (Tc). The multiple
variable regression analysis illustrated in Table 4.1 indicates that Nb
with a p-value of 4.72474E-07 (much lower than 0.05) IS a significant
factor affecting the total cost of printing labels.

As displayed in Tables 4.2a through 4.2d, Nb remains significant
under all the scenarios tested for sensitivity of the model. This
indicates that for all of the value ranges tested, the variable Nb depicts
statistical significance in affecting the total cost of printing labels. In
addition, as seen in Figure 4.5, digital printing becomes increasingly
favorable with respect to conventional printing as the number of

brands increases.

98

Act >0o_o:_._o03 0:35.: .m:o3:0>:oo m:m..0> 23.06 05m:
2035 0:35.: .o “moo :38 00 8:20.53 0.3 8 800mm. .33.. $20 mug-5 .o .0253: .o m:o3m>..0m:o 000;

 

 

 

 

(IauonuaAuoo-Iaufilo)
zsoo mo; :0 aouelema

 

0:355 .0204 “—0 «000 .030... :o 32 “—0 «009:.

 

 

 

055...... .090.— 00 umou .30... :o 03:95 no 500532 no #0095

mJV mun—UH"—

99

EFFECT OF NUMBER OF GLOBAL MARKETS

Figure 4.6a illustrates the one thousand observations of the
independent variable number of global markets (N...) with respect to
the dependent variable the difference of total cost of printing labels
using digital versus conventional printing technology (Tc). The multiple
variable regression analysis illustrated in Table 4.1 indicates that Nm
with a p-value of 2.83556E-90 (much lower than 0.05) IS a significant
factor affecting the total cost of printing labels.

Similar to Nb, Nm remains significant under the business
scenarios tested for sensitivity of the model. This indicates that Nm is
effective in the scenarios tested for sensitivity of the model. This
designates that for all of the value ranges tested, the variable Nm
depicts statistical significance in affecting the total cost of printing
labels. However, distribution does not display a strong trend such as
the Nb plot does (i.e., digital printing becomes increasingly favorable
with respect to conventional printing as the number of brands
increases). This is due to the range of values (i.e., 1, 3, and 4) taken
as representative for Nm. When a product is marketed to a range of
one, 15, and 20 global markets for example, the trend becomes
pronounced as seen in Figure 4.6b (i.e., digital printing becomes

increasingly favorable as the number of global markets increases).

100

6...... >0o.o::o03 0:35.... .m:o3:0>:oo m:m..0> .0506 05m: 2000.

0:35.: .o #60 .38 “.0 8:20.50 05 8 800mm: 5.3 0:2. 39:05 .0020 .o .0053: .o m:o3m>..0mno 000;

 

8°68»-
- - . . 8°52»-
.- » 08.88-
N 80.88-
-l~ ¢ . . 8°85.
m 6090 - - 000.com”-
A“... «l- 8°88-
.m «0: Jo. 80.8.»-
o f 0%

 

Iauonuanuoo-MIBICI)
:soo mo; Io eoualema

 

000.00 5
m. w 000.00N$

0:35.... .003 no .000 .50... :o 82 00 800E.

 

 

 

0505.5 .000.— 00 0000 .000-r :0 300.50.). .0020 no 0005:: no 0009:”

00.? WKDUHH.

101

.1... >0o.o::o0. 0:355 .0:o3:0>:oo m:m.0> .0506 050: 0.000.

0:35.: .o .08 .008 .0 050.050 003 o. .0030. 00.; A520 0.0305 .0020 .o .0053: .0 503030000 000;

 

 

 

 

 

 

:2
80.80..»- a
80.08.;- m m
.m. m
80.88- m .u.
.I 0
28.88. M0 n
- 08.8%- m m

U
08.800- W M
u m.
om W W

80.88
0:35... .0000 .o «000 .030... :o 52 .o .0005.

 

0N 3:0 .0." .H 3 30000.05 m0:.0> no 000:0. 5:
0505.0 .0000 no 0000 .000... :0 0005.02 .0020 no .00532 .0 0009:“—

nwé 9:305.

102

EFFECT OF NUMBER OF ANNUAL PRODUCTION RUNS

Figure 4.7 illustrates the one thousand observations of the
independent variable number of annual production runs (Npr) with
respect to the dependent variable the difference of total cost of
printing labels using digital versus conventional printing technology
(Tc). The multiple variable regression analysis illustrated in Table 4.1
indicates that Npr with a p-value of 1.71365-250 (much lower than
0.05) IS a significant factor affecting the total cost of printing labels.

Similar to Nb and Nm, Npr remains significant under all the
scenarios tested for sensitivity of the model. This indicates that for all
of the value ranges tested, the variable Npr depicts statistical
significance in affecting the total cost of printing labels. As the number
of annual production runs increases, the chances of being able to
incorporate any regulatory change, marketing change, or packaging
change into the already scheduled production changeovers increase.
Therefore, the greater the number of annual production runs, the
easier it is for companies to accommodate changes such as regulatory
ones with less impact on total cost. Similar to the analysis on Nb,
digital printing becomes increasingly favorable with respect to
conventional printing as the number of annual production runs

increases.

103

.35 30.05.00. 0:355 .0:o3:0>:0o 030.0> .0503 0503 0.000. 0:355
.0 .000 .030. .0 050.053 0.3 0. 300000. .35. .52. 0:3. :0303305 .0350 .0 .0053: .0 0:030>.000o 000.0

 

08.800-
000.00.0-
08.800-
08.800-
000.0000.
000.0000-
000.0000.
.7. 08.8.0.
,. ow

|5 000.080
5 000.800

 

 

 

 

.0- .._.7 ,

 

 

(Iauonuenuoo-Ieufilo)
1803 IBIOL 1° eouaJama

 

 

00.00.... .003 .0 .000 .800 co .02 .0 .80....

 

 

0505.... .0000 no 0000 .000... :0 0:3. :0.uu:30.n. .0:::< .0 .0053: .0 0000::

hi NKDUHH.

104

MULTIPLE VARIABLE REGRESSION ANALYSIS

Four of the seven product factors are proven to have an impact
on the total cost of printing package labels in the pharmaceutical
industry. These four are the frequency of marketing based design
changes (ch), number of brands (Nb), number of global markets (Nm),
and number of annual production runs (Npr). Table 4.1 illustrates the
p-values of each of the seven independent variables and annual
average volume (V).

In order to examine the contribution of each independent
variable to the regression model, backward elimination method is used
as part of the sequential search method under the multiple variable
regression analysis framework (Hair, et. al., 1998). This procedure
computes a regression analysis with all the independent variables and
then deletes the independent variables that do not contribute
significantly (Hair, et. al., 1998). Therefore, multiple regression
analysis is exercised continuously by taking the independent variable
with the highest p—value out of the simulation system and running the
multiple regression analysis with the remaining independent variables.
This process is repeated until there are no independent variables left

with a p-value higher than 0.05. The independent variables are taken

105

out of the regression analysis in the following order: Vp, Frc, fix, and
finally V.

Consequently, four independent variables (i.e., ch, Nb, Nm, and
Np.) prove to have a statistically significant impact on the total cost of
package label printing. Tables 4.4 through 4.7 illustrate the multiple
regression analyses following the procedure described above. Table
4.1 displays the regression results for all independent variables tested
and highlights the four independent variables that have a statistically
significant effect on the total cost of printing labels. The resulting
regression analysis is displayed in Table 4.3, where Vp, Frc, ch, and V
are removed from the model confirming the statistical validity of the

final set of independent variables ch, Nb, Nm, and Npr.

TABLE 4.3
Multiple Variable Regression Analysis:

Independent Variables Taken out of the Analysis: Vp, Frc, ch, and V

 

Regression Statistics

 

 

 

R Square .727

Coefficients t Stat P-value
Intercept 221921.236 15.401 .000
Pm -7515.277 -3.231 .001
Nb -14793.221 -22.539 .000
Nm -15814.916 -5.039 .000
Npr -8758.064 -46.412 .000

106

EXAMINATION OF RELATIVE IMPORTANCE

OF PRODUCT FACTORS

The Cohen, Cohen, West, and Aiken (2003) procedure was used
to test the differences between the four beta coefficients (i.e., ch, Nb,
Nm, and Np.) in the multiple regression equation that were found to be
significant predictors on the criterion (dependent) variable, namely the
difference of total cost of printing labels digitally versus conventionally
(Tc). In this case, the standard error of the difference between the

beta coefficients 8- - Bj is given by

 

SE =\/1;R..2y'_‘2_“’i.(rii +r” —2r'7) (4 1)
fifi n—k-l ' '

The observed difference is then given by

zfli —flj

4.2
SE” ( >

t

with the degrees of freedom (df) = n-k-l.

Table 4.4 displays the inverse of the correlation matrix for each
of the product factors and the dependent variable, namely the

difference of total cost of printing labels digitally versus conventionally

(Tc)-

107

TABLE 4.4

Inverse of the Correlation Matrix

Frc ch ch vp Nb Nm Npr Tc
Fm 1.003
Fm -.007 1.014
F... .011 -.017 1.001
v, -.002 .013 .010 1.003
N, .015 .022 .006 .056 1.512
N"1 .016 .038 -.002 .015 .133 1.026
Npr .081 .168 —.034 .091 1.076 .204 3.182
Tc .063 .191 -.014 .082 1.363 .304 2.822 3.657

Six comparisons were assessed (i.e., ch vs. Nb, ch vs. Nm, ch
vs. Npr, Nb vs. Nm, Nb vs. Npr, and Nm vs. Np.) using the inverse of the
correlation matrix and the regression results as input. We found
differences between ch vs. Nb (t-value = 12.28, df = 992, p < .01),
ch vs. Npr (t-value = 22.091, df = 992, p < .01), Nb vs. Nm (t-value =
11.598, df = 992, p < .01), Nb vs. Npr (t-value = 15.085, df = 992, p
< .01), and Nm vs. Npr (t-value = 21.511, df = 992, p < .01), but no
such difference between ch vs. Nm (t-value = 1.652, df = 992). Thus,
the four significant predictor variables can be “ranked” in order of
importance as (Frc, ch, and Vp were not significant predictors in the

regression equation and were thus excluded form the Cohen et al.

test):
Ranking
Number 1: Npr
Number 2: Nb

Number 3 (Tied): Nm
Number 3 (Tied): ch

108

Table 4.5 demonstrates the results of the regression analysis for

each of the product factors.

TABLE 4.5

Results of the Regression Analysis:

Tc Digital-Conventional Criterion Variable

Predictors [3

Fr: -.017
ch —.052
ch .004
Vp -.022
N, -.373
Nm -.083
N...- -.772

8 Std. Error
-3973.897 3846.175
-6912.782 2206.031

907.415 4008.207
6591.410 4892.573
14707-210 656.657
15725.395 3144.292
8788.041 189.378

t-value

-1.033
-3.134
.226
-1.347
-22.397
-5.001
-46.405

R2 = .727, Adjusted R2 = .725. 1B = standardized beta and “b” =
unstandardized beta. R2 is the total R2 for the equation. Adjusted R2 is the
total R2 for the equation taking into account the number of predictor

variables and the sample size.

Sign (p)
.302
.002
.821
.178
.000
.000
.000

The next chapter presents the results of the data analysis. Each

of the hypotheses is reviewed and the outcomes are revealed with

explanations in which digital versus conventional methods of printing

are favored with respect to each scenario.

109

CHAPTER V

RESULTS

This study set out to analyze package labeling cost structure in
the implementation of postponement strategies. First, digital printing
is identified as an enabler to postponement. Then, product factors that
affect the decision to implement postponement strategies are
revealed. Next, the typical value ranges for these product factors are
identified through case study interviews. The effect of each factor on
the total cost for label printing and the decision to postpone is
analyzed. Finally, the product factors that have the most significant
impact on the total cost for label printing and the decision to postpone
label printing to the distribution center level are identified.

In this chapter, the effect of each product factor on the total cost
of package label printing for the pharmaceutical industry is analyzed
and the results are presented. The combination simulation-regression
analysis conducted on the total cost model represents the value ranges
of the pharmaceutical industry data. The one thousand observations
create a large sample size and help generalize the results. Four of the
seven product factors are proven to be effective on the total cost of
printing package labels in the pharmaceutical industry. It is also

observed that in the majority of the one thousand observations, digital

110

printing proved to be advantageous compared to conventional printing
from a total cost perspective.

The product factors that are found to have a significant effect on
the total cost of printing labels are ch, Nb, Nm, and Npr; and F.c and ch
are also found to have a significant effect on the total cost of printing
package labels when the range of values are increased. Employing the
simulation model on their respective data enables companies to assess
whether or not they should print using digital technologies and
therefore postpone labeling to the distribution center level. In other
words, when minimum, mode, and maximum ranges of values are set
into the model, the model suggests a type of printing technology to be
used (i.e., digital or conventional) based on the total cost of package
label printing. As seen in Figures 4.1 through 4.7, the representative
ranges of values used in this study favor digital printing methods to be
used over conventional methods.

Next, the hypotheses for this research are presented with the

observations derived from the model.

EFFECT OF FREQUENCY OF REGULATORY CHANGES

A. The HIGHER the frequency of regulatory changes (Frc), the

GREATER the incentive to postpone labeling to the distribution

111

center level, and the GREATER the cost of obsolescence and the
cost of printing runs (most upgrades involve more information to be

put on labels).

With the representative value ranges presented in Chapter III, F..-_ is
rendered insignificant in terms of its effect on the total cost of
package label printing. However, as the value ranges are increased,
the significance of this factor surfaces. As Figure 4.1 indicates, a
large majority of the one thousand iterations on the simulation
model favor digital printing over conventional printing. As indicated
previously, digital printing is an enabler to postponement
strategies, and therefore there is a greater incentive to postpone

labeling to the distribution center level.
EFFECT OF FREQUENCY OF MARKETING CHANGES
B. The HIGHER the frequency of marketing based changes (ch). the
GREATER the cost of obsolescence and the cost of printing runs

(most changes involve more color to be put on labels; or more

expensive colorant, such as metallic ink).

112

With the representative value ranges presented in Chapter III, ch
is found to be significant in terms of its effect on the total cost of
package label printing. As Figure 4.2 indicates, a large majority of
the one thousand iterations on the simulation model favor digital
printing over conventional printing. Therefore, as ch increases,
there is a greater incentive to postpone labeling to the distribution

center level.

EFFECT OF FREQUENCY OF PACKAGE ENGINEERING CHANGES

C. The HIGHER the frequency of packaging engineering design
changes (Ppc), the GREATER the cost of obsolescence and the cost
of printing runs (most changes involve more complicated designs

such as folding to better prevent shock and vibration damage).

Similar to Frc, ch is rendered insignificant in terms of its effect on
the total cost of package label printing. However, as the value
ranges are increased, the significance of this factor is revealed. As
Figure 4.3 indicates, a large majority of the one thousand iterations
on the simulation model favor digital printing over conventional

printing. As noted previously, digital printing is an enabler to

113

postponement strategies, and therefore there is a greater incentive

to postpone labeling to the distribution center level.

EFFECT OF VALUE OF PRODUCT-PACKAGE SYSTEM

D. The GREATER the value of product—package system (Vp), the

GREATER the incentive to postpone labeling to the distribution

 

center level (generally, as the value of product-package system

increases, the cost of printing labels increases).

Similar to Frc and ch, Vp is rendered insignificant in terms of its
effect on the total cost of package label printing. However, as the
value ranges are increased to the levels of 5, 500, and 10,000, the
significance of this factor surfaced. As Figure 4.4a indicates, a large
majority of the one thousand iterations on the simulation model
favor digital printing over conventional printing with the
representative value ranges gathered through the case study
interviews initially. However, in the case of very specialized
pharmaceutical treatments, such as cancer, or in the case of
surgical implants, the Vp ranges could be very high. The second

simulation observed the effect of those cases. As demonstrated in

114

Figure 4.4b, it is very hard to suggest one printing method over the

other. Therefore, this hypothesis does not hold true.

EFFECT OF NUMBER OF BRANDS

E. The GREATER the number of brands (Nb), the GREATER the

incentive to postpone labeling to the distribution center level.

Similarly, Nb is found to be significant in terms of its effect on the
total cost of package label printing. As Figure 4.5 indicates, a large
majority of the one thousand iterations on the simulation model
favor digital printing over conventional printing. Therefore, as Nb
increases, there is a greater incentive to postpone labeling to the

distribution center level.

EFFECT OF NUMBER OF GLOBAL MARKETS

F. The GREATER the number of global markets (Nm), the GREATER the

incentive to postpone labeling to the distribution center level.

Nm renders to be significant in terms of its effect on the total cost of

package label printing. As Figure 4.6 indicates, a large majority of

115

the one thousand iterations on the simulation model favor digital
printing over conventional printing. Therefore, as Nm increases,
there is a greater incentive to postpone labeling to the distribution

center level.

EFFECT OF NUMBER OF ANNUAL PRODUCTION RUNS

G. The GREATER the number of annual production runs (Npr), the
LOWER the incentive to postpone labeling to the distribution center

level.

Npr is found to be significant in terms of its effect on the total cost
of package label printing. As Figure 4.7 indicates, a large majority
of the one thousand iterations on the simulation model favor digital
printing over conventional printing. Therefore, as Npr increases,
there is a GREATER incentive to postpone labeling to the
distribution center level. This is probably the result of digital
printing supporting more production changeovers compared to
conventional printing, from a total cost perspective. Therefore, the
relationship found between Npr and the dependent variable is in the
OPPOSITE direction of the hypothesis, hence the hypothesis has

been disproved.

116

CONCLUSION

The number of annual production runs (Npr) is the number one
factor impacting the difference of total cost of printing labels digitally
versus conventionally. Number of brands (Nb) is the second factor
impacting the difference of total cost of printing labels digitally versus
conventionally. Finally, number of global markets (N...) and frequency
of marketing based graphic design changes (ch) are tied up as the
third factors.

Three of these factors, namely Nb, Nm, and ch, stem from
strategy-based managerial decisions. The greatest impact is attained
by Npr, which suggests that the decision of running productions weekly
versus monthly has significant bearing on total cost of printing labels.
In addition, the pronounced trend line observed from the simulation
results indicate that the cost savings realized through digital printing
increases progressively as the production schedule moves from for

example, monthly to weekly.

117

CHAPTER VI

DISCUSSION AND IMPLICATIONS

This chapter discusses the implications of this research and the
areas where further research is needed. In light of the results
presented in Chapter V, this study develops a total cost model that
succeeds in generalizing the different scenarios when different printing
technologies could be employed and postponement strategies
implemented.

For a blockbuster drug (over $1 billion in sales per year), a one-
day delay of the initial launch of a new product costs up to $3 million
(Trombley, 2003). The difference of timing may be as much as a
couple of days if the labels are printed digitally. So, digital printing
could possibly save a pharmaceutical company as much as $6 million
in the beginning of a product launch (Trombley, 2003).

This study also develops a simulation model based on total cost
considerations for package label printing for the pharmaceutical
industry. The model identifies the product factors that have a
statistically significant impact on the total cost for printing labels and
compares digital versus conventional printing methods. The simulation
model reveals the product factors that have a significant affect on total

cost of package label printing. When minimum, mode, and maximum

118

 

ranges of values are put into the model, the model suggests a type of
printing technology to be used (i.e., digital or conventional) based on
the total cost of package label printing. The representative ranges of
values used in this study favors digital printing methods to be used

over conventional methods.

RESEARCH LIMITATIONS

From a business perspective, the motivation to postpone is to
maximize profits primarily through inventory reduction. In a general
sense, profits equal revenues minus total cost. For the purposes of this
study, revenues are assumed constant. However, it is also important
to recognize the fact that postponement also has effects on revenue.
For example, due to the simple fact that by utilizing postponement
strategies, companies are able to customize certain products according
to customers’ needs, and therefore, have the customer pay more for
the product than a generic counterpart.

From the pharmaceutical industry point of view, the revenue
gain will be realized through faster launch of patented products once
approvals are received. The subject under investigation,
postponement-speculation, has diverse effects on the speed of launch
of a particular product. For example, assuming that the label wording

will need to be changed per FDA approval, a company that postpones

119

label printing until product approvals are received may actually
shorten the product’s time to market. On the other hand, a disastrous
situation may occur when labels are printed prior to receiving the
approvals with a speculative approach on the label requirements and
then having the labels and/or inserts re-printed with accurate
information.

Another business scenario is to print the labels with speculative
wording and actually end up gaining time in launching the product if
the approval is received without any changes in label wording. There
are situations where pharmaceutical companies make such decisions
based on experience. The case where materials are printed before
approval is described as “at risk,” meaning that if there are any
changes in the wording for the labeling, the printed materials have to
be destroyed (McNeely, 2001). In addition, McNeely (2001) notes that
it is sometimes advisable to print the inserts “at risk," since it becomes
a rate-limiting step in the launch process.

However, there should be some cases where not taking such
risks may also depend on the company having digital printing
capabilities that boast minimal setup time, and this alone may justify
the costs associated with labeling postponement. On the other hand,
for this research, revenue is assumed to be constant, and therefore

maximizing profit will depend solely on minimized total cost. At the

120

same time, the study recognizes the importance of analyzing the
effects of postponement on revenues, both from a purely research

contribution as well as from a business perspective.

FUTURE RESEARCH CONSIDERATIONS

As explained, for the purposes of this study the revenue function
is assumed to be constant. The total cost of printing labels is a
function of the seven product factors identified in this research and
postponement, and total cost is assumed to be a continuous function
with respect to postponement. Therefore, the study assumes that in
order to maximize profits, total cost should be minimized. However,
profits equal revenue minus total cost. Thus, it would be interesting to
research the ways that the seven product factors (or any additional
factor) affect the profit function by looking at how the profit function
could be maximized.

The questions this research addresses are: What factors need to
be considered in evaluating effectiveness of postponement? Under
what circumstances are new technologies such as digital printing cost
effective? Is there a model that could be developed to bring answers
to these questions? Issues such as profitability, supply chain visibility

and integration concepts have not been addressed. Customers demand

121

 

more supply chain visibility, as indicated by Richardson (2000). Since
printers can increase visibility into the packaging process (Richardson,
2000), could supply chain visibility be one of the efficiencies gained?
For achieving increased production rates and information availability,
printer applications are integrated into warehouse management
systems (WMS), enterprise resource planning (ERP), and order
processing systems (Richardson, 2001). What would be the effects of
postponement on integration of printer applications into WMS and ERP
for increased production rates and supply chain visibility? Addressing
these questions in one or more research projects would contribute
greatly to the scholarly puzzle of packaging supply chain.

As the printing technologies advance, the dynamics of the supply
chain could change. For example, recent developments in lithographic
printing technology include utilizing computer to plate processing. This
advancement reduces setup time (and costs) however, currently ends
up costing more (Ingram, 2002). Nevertheless, as technological
advancements occur, there is great potential for the cost model for
package label printing to change. Further research updates are
necessary in order to document such alterations.

Another area that warrants further research is the effect of
product tracking technologies on the total cost of package label

printing. Barcode and, more recently, radio frequency identification

122

 

(RFID) technologies have revolutionized the supply chain by providing
on-time and accurate information regarding the product; in some
cases including the temperatures that the product has been exposed
to. Through RFID, the quality of information provided varies
tremendously and the quantity increases as the technological
advancements take place at ever increasing rates. This not only affects
the cost of printing labels, but may also dictate certain changes in the
supply chain for printing labels and packaging.

Finally, additional research could be conducted for different
stages of the pharmaceutical product life cycle. For example, the
ranges of values for the product factor frequency of regulatory
changes vary depending on where the product is in its life cycle. In the
first year of the patent, the regulatory changes tend to be much more
frequent. Similarly, some of the cost components may change
depending on where the product is in its life cycle. Therefore,
developing a total cost model of pharmaceutical products by observing
the different stages of the product life cycle may uncover diverse

results.

123

MANAGERIAL IMPLICATIONS

There remains one last question to be addressed: What are the
implications of this total cost simulation model for management? The
model creates a tool for evaluating the economic impact of digital
versus conventional printing for a particular product line in the
pharmaceutical industry. This study treats digital printing as an
enabler to postponement, therefore a negative cost difference between
printing labels digitally versus conventionally indicates cost savings
realized through digital printing, and hence incentive for
postponement. Consequently, the research establishes the groundwork
under which postponement becomes viable from a total cost savings
perspective. In addition, the total cost model suggests using one
printing method over the other (i.e., digital versus conventional) based
on the economical justification. The analyses of the following business
situations demonstrate the generalizations with respect to
management decisions under which digital printing (i.e.,
postponement) is economically justified.

Four of the seven product factors are proven to be effective on
total cost savings achieved through digital printing technology for
printing package labels in the pharmaceutical industry. As a result,

only one of the hypotheses (i.e., the effect of value of the product—

124

package system) is proven inconclusive, and one was disproved (i.e.,
found to be in the opposite direction of the hypothesis relating number
of annual production runs to the total cost model). The product factors
that are found to have a significant effect on the total cost savings
realized through digital printing technology for printing labels are
frequency of marketing changes (ch), number of brands (Nb), number
of global markets (Nm), and annual number of production runs (Npr);
and frequency of restriction changes (Frc) and frequency of package
engineering design changes (ch) when the range of values are
increased.

In addition, it is observed that for the majority of the one
thousand observations created by the minimum, mode, and maximum
representative values, digital printing proved to save total cost
compared to conventional printing.

Total cost savings achieved through the use of digital printing
becomes significant if a particular pharmaceutical product label goes
through regulatory changes (F...) on average three times a year, with a
range of one to five times. Based on this study, this also indicates that
the decision to postpone labeling to the distribution center level is
economically justified. However, if the Frc values are between one and

three and average at two for that product, then the cost saving is not

125

conclusive. As the frequency of regulatory changes increase annually,
the significance on the effect of cost savings continues to hold.

As presented in Chapter IV, ch is an important variable in the
decision to postpone labeling to the distribution center level, when
marketing changes occur one to five and average at three times a
year. When the representative values for Frc are increased to equal the
values of ch, both become effective factors in the decision to
postpone. Consequently, the frequency of marketing based design
changes a product label goes through affects the cost savings realized
through using digital printing technologies. As ch increases, greater
cost savings are achieved through the use of digital printing
technologies, and therefore, postponing labeling to the distribution
center level.

Similar to Frc, ch becomes a significant factor affecting the total
cost of printing labels, hence the decision to postpone labeling, when
the representative values are increased to equal ch values (i.e., one
to five and average at three). Therefore, under the business
circumstances presented previously, postponing package labeling to
the distribution center is justified when the number of annual package
engineering changes vary between one and five and average at three.

When the simulation model is run holding everything constant

and varying the value of product-package system (Vp) values between

126

$5, $500, and $10,000, thedifference of cost between digital and
conventional printing appear evenly distributed (Figure 4.4.b). Hence,
in the case of high-end pharmaceutical products such as surgical
implants, total cost and/or postponement implications for label printing
is inconclusive. Similarly, other values that are tested with this model,
proved the value of product-package system to be inconclusive in
terms of its effect on realizing total cost savings through the
employment of digital printing technologies, and postponing labeling to
the distribution center level in the supply chain.

The variable Nb depicts statistical significance in affecting the
total cost of printing labels. As the number of brands increase, digital
printing becomes increasingly favorable with respect to conventional
printing. The model is tested for the number of brands ranging from
five to 20 and averaging at 12 and demonstrates that the factor is an
important variable in the postponement decision.

Similarly, the variable Nm portrays significance in the decision to
postpone. If a product line is marketed in one to 20 global markets,
digital printing becomes increasingly favorable from a total cost
perspective compared to conventional printing, as the number of
global markets increase. Consequently, total cost savings achieved
through digital printing technologies become increasingly greater if a

particular pharmaceutical product line is marketed in 15 global

127

markets on average, and ranging from one to 20. This indicates that in
such cases, postponing the labeling function to the distribution center
level is economically justified.

Npr depicts statistical significance in affecting the difference of
total cost between digital and conventional printing, hence the decision
to postpone labeling to the distribution center level. As the number of
annual production runs increases, the chances of being able to
incorporate any regulatory change, marketing change, or packaging
change into the already scheduled production changeovers increase.
Therefore, the greater the number of annual production runs, the
easier it is for companies to accommodate changes, such as
regulatory. In addition, this relationship also indicates that on the
contrary to what was hypothesized, the inventory savings turned out
to be greater than stock-out costs. Similar to the analysis on Nb, digital
printing becomes increasingly favorable with respect to conventional
printing as the number of brands increases.

The relative importance of the independent variables that are
found to be statistically significant in impacting the difference of total
cost of printing labels digitally versus conventionally is also analyzed.
The results indicate that the number of annual production runs is the
number one factor in terms of its importance. Therefore, production

scheduling is the biggest factor that management should look at in

128

making the decision to print labels digitally versus conventionally. In
addition, the simulation model demonstrates that as the production
schedule moves from, for example, monthly to weekly, cost savings
realized using digital printing becomes increasingly more. The second
factor that is most important from a cost savings perspective is the
number of global brands. Cost savings realized through digital printing
technologies increase progressively more, as the number of brands
increase for a particular compound. Similarly, the number of global
markets a pharmaceutical product is marketed in, as well as the
frequency of marketing based design changes that a label for a
particular product is exposed to, impact the total cost savings realized
through using plateless printing technologies.

The four cost drivers identified in this study are printing run,
inventory, stock-out, and obsolescence. The cost drivers that are
shown to have a significant impact on the total cost savings achieved
through the use of digital printing are printing run, inventory, and
obsolescence. The variable, number of annual production runs, also
indicates that the inventory cost savings realized through plateless
printing are higher than the cost of stock-outs.

The pharmaceutical industry could benefit tremendously from
further research in the aforementioned matters. Several smaller scale

research topics could cover different niche market segments within the

129

pharmaceutical industry, such as over-the-counter drugs, surgical
implants, medical equipments, and prescription drugs in more detail.
Similarly, additional investigations could be conducted to research
other industries, such as food and beverage industry.

As a conclusion, this study paves the way for research
opportunities to serve the scholars, businesses, and government
agencies in identifying factors that affect total cost and profit functions
of package label printing for diverse industries. Addressing these
factors would greatly benefit the businesses and academia in

respective industries.

130

 

APPENDIX A

INTERVIEW QUESTIONS

Ques'rlous REGARDING PRODUCT FAggks
1. Frequency of Regulatory Changes (Fm)

 

[Frc Minimum: Mode: Maximum:

Comments:

2. Frequency of Marketing Based Design Changes (ch)

 

[ch Minimum: Mode: Maximum:

 

Comments:

3. Frequency of Package Engineering Design Changes (ch)

 

[ch Minimum: Mode: Maximum:

 

Comments:

4. Value of Product-Package System (Vp)

 

 

 

 

[Vp Minimum: Mode: Maximum:
Np (digital) Minimum: Mode: Maximum:
Comments:

5. Number of Brands (Nb)

 

FNbMinimum: Mode: Maximum:

 

131

Comments (different range of values if the labels are printed

digitally?):

Number of Global Markets (Nm)

 

[Nm Minimum: Mode: Maximum:

Comments (different range of values if the labels are printed

digitally?):

Number of Annual Production Runs (Npr)

 

[Npr Minimum: Mode: Maximum:

 

[Npr (digital) Minimum: Mode: Maximum:

Comments:

Comments about product factors in general

132

E TI RE ARDIN T FA
c=PR+INV+STO+OBS

R

1. Cost of Printing Run (PR)

CO = Average Annual Order Quantity

V = Average annual volume = OQ x (Npr /2)

PR = PS + CONSTPR X V
PS = Nb X Nm X CONSTps

PR = (Nb x N... x CONSTPs) + (CONSTPR x V)

= (Nb x Nm x CONSTps) + (CONSTpR x (OQ x (Npr /2)))

CONSTpR = Annual expenses on substrate usage, colorant usage, parts
replacement, etc. divided by the #of annual setup changes. CONSTpR
has two different values for the case of digital and conventional
printing, $7 per day for conventional and $30 per day for digital

(Emden, 2001-2003).

CONSTps = Annual expenses on labor and materials for setup (double
sided tape in the case of conventional printing and color setup mostly

in the form of labor in digital)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[V Minimum: Mode: Maximum:
[V (digital) Minimum: Mode: Maximum:
IOQ Minimum: Mode: Maximum:
IOQ (digital)Minimum: Mode: Maximum:
ICONSTpR Minimum: Mode: Maximum:
[CONSTPR (digital) Minimum: Mode: Maximum:
[CONSTps Minimum: Mode: Maximum:
ICONSTps (digital) Minimum: Mode: Maximum:

 

Comments:

133

2. Cost of Inventory (INV)

INV = Product-Package System Value (Price) = Vp x CONSTICC x V

SS = Safety Stock

CONSTICC = Inventory carrying cost percentage factor (constant)
CONSTICC = 1.5% (time value of money tied up in inventory on a
monthly basis)

Unit cost of printing a single label = UT

INV= (((V/ Np.) + SS) x u. x CONSTps) x CONSTICC
= (((OQ/ 2) + SS) X UT X CONSTPS) X CONSTICC

 

 

 

 

 

 

 

 

 

 

[CONSTICC Minimum: Mode: Maximum: J
[UT Minimum: Mode: Maximum: J
[UT (digital) Minimum: Mode: Maximum: I
LSS Minimum: Mode: Maximum: I
[SS (digital) Minimum: Mode: Maximum: j
Comments:

3. Cost of Stock-out (STO)

STO = CONSTSTO x (100 - SL) x Mp

Mp = Profit Margin

SL = 1 - (f(k) x CONSTO) / OQ

CONSTO = Volume replenishment lead time

If SL = 98% then STO = (1 - 0.98) x V

Demand Standard Deviation (OQSD)

Ordering Cost (OC): Including purchasing and receiving (inspection,
etc.)

STOpd = Stock-out cost per day

 

[SL Minimum: Mode: Maximum: 7

 

lOQso Minimum: Mode: Maximum:

134

 

[OC Minimum: Mode: Maximum:

 

 

 

 

[OC (digital) Minimum: Mode: Maximum:
ETOpd Minimum: Mode: Maximum:
Comments:

4. Cost of Obsolescence (OBS)

CONSTobS = Scrap Handling / Disposal Cost Constant

 

[CONSTc-bS Minimum: Mode: Maximum:

 

Comments:

5. Error Term (Projection versus Actual Cost)
Typically, what’s the range of percentage difference between your

projected costs (budget) and actual costs (accrued expenses)?

 

[Error Term Minimum: Mode: Maximum:

 

Comments:

6. Comments about cost factors in general

135

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.0 0 F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F N
.0 0 F 0N NF 0 N F 0.0 N F 0 0 0 F N N F N
.0 0 F 0N NF 0 N F 0.0 F 0 m 0 F 0 N F N
.0 N F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F NN
.0 0 F 0N NF 0 N F 0.0 N F 0 0 N F N N F FN
.0 0 F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F 0N
.F. 0 F 0N F 0 N F 0.0 N F 0 0 0 F 0 N F 0F
.0 0 F 0N F 0 N F 0.0 N F 0 0 0 F 0 N F 0F
.4 0 F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F E
.0 0 F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F F
.v 0 F 0N NF 0 F 0.0 N F 0 0 0 F 0 N F F
.0 0 F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F F
.0 0 F 0N NF 0 N F 0.0 N F 0 0 F 0 N F 0F
.F. N F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F F
.0 0 F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F FF
.F. 0 F 0N NF N F 0.0 N F 0 0 0 F 0 F S
.0 0 F 0N F N F 0.0 N F 0 0 0 F 0 N F 0
.0 0 F 0N F 0 N F 0.0 N F 0 0 N F 0 N F 0
.0 0 F 0N NF 0 N F 0.0 N F 0 0 0 F N N F R
.0 0 F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F 0
.0 0 F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F 0
.0 0 F 0N NF N F 0.0 N F 0 0 0 F 0 N F 3
.0 0 F 0N NF 0 N F 0.0 N F 0 0 0 F 0 N F 0
.0 N F 0N NF 0 N F 0.0 N F 0 m 0 F 0 F N
7m. 0 F 0N NF N F 0.0 N F 0 0 F N N F F

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.55 0305 55 .55 0305 55 .55 0305 55 .85 0305 55 .85 0305 55 .05 0305 5E
52 52 52 02 02 02 0> 0> 0> 00... 00“. 00.. 05". 05“. 05“. 0.“. 0.". 0.“.

3003.00. :003 0.00..0>0 0.0 .0305 530.350 03 3:0 0.03 :3“. 00305
530.350 0:. 50... 0.03 0:. .0 0.0500 _.050 0 0500050. >50 05.0 .000; 002. .0305 0.3 :0 :3. 0:030>.0000 .0 .0053: .030...

nan—OZ ZOHPSDEHM ZH mZOHP<>¢mmmO mN hwy—H".
m XHDZmna<

136

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.0.NON.. NON.00..0N .. .0 0. N .00 N N 000000.00. 000.000.0N 000.000 00 N. N 0N
000.00N 0N0.000.0. .0N 0 0. . 0.. 0 N 000.000.00. 000.000.0N 000.000 00 N. N 0N
00.000 .00.000.0. .0. 0 .. . 0.. N N 000000.00. 000.000.0N 000.000 00 N. N 0N
000.000 .N0.000.0N NN 0 0. . 0.0 0 N 000000.00. 000.000.0N 000000 00 N. N NN
00.000 000.NON.0N 0. N 0 N 0.. N . 000000.00. 000.000.0N 000.000 00 . N .N
.0N.0NN.. .00.000.00 0N 0 N. . 0.. 0 N 000.000.00. 000.000.0N 000.000 00 N. N 0N
000.000 0N0..00. .0 .00 .0 .. . 0.. N 00000000. 000.000.0N 000.000 00 N. N 0.
000.0N0 00003.00 00 0 0. . N0 0 00000000. 000.000.0N 000.000 00 N. N .
000.000 0.0.N.0.00 00 N. N 0.0 N 0 00000000. 000.000.0N 000.000 00 N. N 0.
00.000 000.000.0N 0. N .. . 0.0 0 000.000.00. 000.000.0N 000.000 00 N. N 0.
.0000. 0.0.000.0 ON 0 0 . 0.0 N N 000.000.00. 000.000.0N 000000 00 N. N 0.
00N.00. 00..000.o. .0 0 0. . N0 0 N 000000.00. 000.000.0N 000.000 00 N. N .0.
.00.000 000.0N0.N0 00 .0 .. . 0.0 0 N 000.000.00. 000.000.0N 000.000 00 N. N .
000.0N0.. 000.000.N0 NN N 0. N 0.0 .0 N 000000.00. 000.000.0N 000.000 00 N. N .
000.0.0 000.000.0N 0N 0 0 N ... N N 000000.00. 000.000.0N 000.000 00 N. N ..
0...000 000.000..N .. 0 0. . 0.0 0 N 000000.00. 000.000.0N 000.000 00 N. N 0.
000.000 000.00N.N. 0. . 0 . 0.. . N 000.000.00. 000.000.0N 000.000 00 N. N 0
N00.00..N 000.000.0N 0 0 0. . .. 0 . 000000.00. 000.000.0N 000.000 00 N. N 0
.00.000 000.000.00 00 . 0. . ... . 0 000.000.00. 000.000.0N 000.000 00 N. N 0
0N0.0.N N.0.0.0.0. 00 0 0 . 0.0 N 000.000.00. 000.000.0N 000.000 00 N. N 0
N0..000 00..000.0N 0. 0 N. . 0.0 0 . 000.000.00. 000.000.0N 000.000 00 N. N 0
0.0.000 000.0000. N. N 0. . 0.. 0 N 00000000. 000.000.0N 000.000 00 N. N .0
000.000.N 0000.000 0. 0 N. . N0 .0 N 000.000.00. 000.000.0N 000000 00 N. N 0
000.000 000000.00 N0 0 0. N 0.. 0 N 000.000.00. 000.000.0N 000.000 00 N. N N
000.00..N 0.000000 0. 0 .0. . .0. 0 N 00.000.00. 000.000.0N 000.000 00 N. N .
00 > .02 52 02 0> 00... 05“. 0.“. x05> 0305> 55> x05 0305 55

.02

.02

.02

137

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8 08 80.: 00:88 80.80 880.8 0000.8 08.030 000.08 30.8 08.8 00
0F 02 80.00 3:88 03208 080.8 0800.8 000.008 08.080 08.20 8:08 am
00 000 08.8 000.88 80.80 0500.8 8000.8 08.008 08.8 80.20 000.8 00
t 000 000.00 0.00.008 800:0 0800.8 8000.8 80.38 08.88 000.20 08.8 8
8 000 000.8 000.080 80.08 080.8 80008 000.080 80.80 80.08 80.8 a
0? 000 05.02 000.008 08.8 00000.8 0800.8 08.008 000.08 00:08 000.8 00
: 80 833 000.008 0:808 0000.8 800.8 80.88 88.080 80.08 80.8 9
00 80 0058 80.008 000.080 808.8 0888 03.008 000.08 000.08 0.8.8 0*
0 000 08.00 08.88 08.008 8088 8008 000.008 000.008 80.08 00.20 t
8 80 50.00 08.020 0.00.000 8000.8 30008 80.88 08.08 80.08 02.8 08
2 SF 000.00 000.008 08.08 00000.8 0300.8 00308 08.8 00.0.0 08.8 0F
8 0..: 8.00 08.88 08.88 0800.8 008.8 8.008 08.58 08.08 80.8 0
: .000 000.0: 08.80 80.88 00000.8 0800.8 0388 F0580 000.08 80.8 0F
8 000 80.5 08.08 80.08 8000.8 080.8 80.88 80.000 08.20 .088 8
0F 08 30.8 03.080 8.08 8088 00000.8 000.88 800.08 08.08 80.8 :
00 .00 080 08.008 08.80 8000.8 00000.8 08.030 000.08 08.8 80.8 2
00 08 80.00 80.08 08.08 800.8 8000.8 000.80 000.08 80.8 80.8 0
00 80 80.8 000.08 08.08 0500.8 3.0008 08.80 08.08 2.0.8 0.0.8 0
0 03; 08.08 80.80 08.020 8000.8 05008 000.800 08.008 8.08 08.8 0
0 000 08.00 80.008 8.50 0500.8 00000.8 80.88 8.008 80.8 08.8 0
00 80 808 000.020 20.000 0500.8 0800.8 08.88 80.000 80.50 000.8 0
8 F00 00:00 08.3.0 000.08 0808 00000.8 03.88 08.08 08.8 80.8 a
00 000 8.08 00.88 08.08 8000.8 8000.8 08.020 000.000 000.20 000.8 0
: F00 80.08 08.008 08.88 8000.8 000008 000.008 8.030 08.08 80.8 N
00 08 80.00? 000.80 80.08 880.8 080.8 000.08 000.000 80.80 08.8 _
05: 30 00058 5:8 E .030 E 5:8 .890 $08 0.. .0090 0.. 5:8 Elam”.
0.26 .>< 000500 2.00 .2 3.8200 .3028 00828 0000200

 

 

138

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

00.00 000.00 50.0 00.00 000.0 00.0000 000.50 0 000.0 000.0 00.000; 000.00 00000.0 0
00.00 000.00 00.0 00.00 000 000.000 000.000 0 000.0 000.0 000.000 000.0 0000.0 00
00.00 000.00 00.0 00.00 000.0 000.000; 000.000 0 000.0 000.0 000.000 000.0 0000.0 00
00.00 000.00 00.0 00.00 00: 000.000.. 000.000 0 000.0 000.0 000.000 000.0 0000.0 00
00.00 000.00 0\00 00.00 000.0 000.0000 000.000 0 000.0 0000.— 000.000 000.0 0000.0 00
00.00 000.00 80 00.00 000.0 000.000; 000.000 0 000.0 000.0 000.000.? 000.00 0000.0 00
n0.00 00:0 00.0 00.00 000.0 05000.0 90.000 0 000.0000; 000.000 000.0 0000.0 00
00.00 000.00 0.0 00.00 00: 000.000; 000.000 0 000.0 000.0 000.000 000.0 500.0 00
00.00 000.00 80 00.00 000.0 000.00: 000.000 0 000.0 000.0 000.000 03.0 0000.0 00
00.00 000.00 00.0 00.00 0:; 000.000; 000.000 0 000.0 000.. 000.000 000.0 0000.0 00
00.00 000.0 00.0 00.00 000 00.000 000.00 0 000.0 000.? 000.000 000.0 0000.0 00
00.00 000.0 00.0 00.00 000 000.000 000.000 0 000.0 000.0 000.000 00; 0000.0 .00
00.00 000.00 00.0 00.00 00: 000.000; 000.000 0 000.0 050 000.000 000.0 000.0 00
00.00 000.00 00.0 00.00 000 000.0000 000.000 0 000.0 000.0 000.000.? 000.00 0000.0 00
00.00 0030 00.0 00.00 000.0 000.000; 000.000 0 000.0 000.0 000.000 000.0 0000.0 2
00.00 000.00 00.0 00.00 000.0 000.000; 000.000 0 000.0 000.0 02.000 000.0 0000.0 0.
00.00 000.00 00.0 00.00 000.. 000.000.. 000.000 0 000.0 000; 000.000 00.0 0000.0 0
00.00 000.00— 00.0 00.00 000.0 000.0000 000.000; 0 000.0 000.0 000.0000 000.00 0000.0 0
00.00 000.00 00.0 00.00 000.0 000.000; 000.000 0 000.0 000.0 3.0.000 000.0 0000.0 0
00.00 000.: 80 00.00 000 000.000 000.000 0 000.0 000.0 000.00 000.0 0000.0 0
00.00 000.00 00.0 00.00 000., 000.0000 000.000 0 000.0 000.? 000.000 000.0 0000.0

00.00 000.00 00.0 00.00 000.0 000.000; 000.000 0 000.0 000.0 000.000 000.0 0000.0

00.00 000.000 00.0 00.00 000.0 50000.0 000.000; 0 000.0 000.. 000.0000 000.00 0000.0

n0.00 000.00 00.0 00.00 000.0 000.000; 000.000 0 000.0 000.. 000.000 000.0 0000.0 0
00.00 000.00 00.0 00.00 000.0 000.0000 00.000; 0 000.0 000; 000.0000 000.00 002.0 0
..0 >2. 8.00200 .0850 000.000 000.000 >2. mm .. 0.: 800 as: 000 2::

x020 gs: 08.503 «8:003 008030 05:85 0.000 0.000
0.5 80.5 05.50
00.0an

Ohm

139

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

000.0000 0.00. 00.0 000.0000 000.000 000.000 000.000 000.000 0. 5 000.0000 00
000.0000 000.000 000.00—0 000.30 000. .05 000.000 000.000 0. 00 00.0000 .00
000.0000 000.000 50.00 00 000.000 000.000 N00. 50 000.000 0. 00 000.00 00 00
09.0000 00. :00 000.0000 000.000 000.0? ..0 000. 000 000.000 0. 00 0.00.0000 0.0
000.0000 000.000 000.0000 000.000 000.000 000.000 000.0 ..0 m. ..0 000.0000 ..0
09.0000 000.000 000.0000 000.000 0.0. 000 000.000 000.000 0. 00 000.0000 00
000.0000 50.000 000.0000 00.3000 050.0000 000.000 000.000 0.00 000. 0000 0..
000.0000 50.0 5 000.0000 000.000 000.00 00 0 F .000 50.0 ..0 m. 00 00.0000 00
000. 0000 000.000 000.050 000.000 05.020 000.000 000.000 0. 5 000.0000 00
000.0000 000.000 00.0000 000.000 000.000 00. 000 000.000 0. 5 000.0000 0..
000. 00. ..0 000.0 ..0 20.000 0 00.00 00.000 000.000 000.0 00 m. 00 2.0.000 0
50.9.00 0.0.05 000.000 000.00 000.00 00 000.000 05.05 0.00 000.000 0
000.0 000 000.000 0 00.0000 000.000 000.00 ..0 000.000 000.000 0. ..0 .. 00.0000 9
000.0000 000.000 000.0000 000.000 000.000 000.000 000.000 0. 5 000.0000 00
000.0 00.0 000.0 00 50.0000 000. 000 000.000 000.000 000.0 00 m. 00 000.0000 2.
0 00.0000 000.000 000.0000 000.000 0 00.000 000. .00 000.000 0. ..0 0.00.0000 or
000.0000 000.000 000.0—F0 000.000 000.000 630.000 30.000 0.00 000.030 0
000.0000 000.000 000. .000 50.0000 000.000 000.000 000.000 0.00 000. .000 0
000.0000 000.0 00 05.0000 000.000 000.00 ..0 000.000 000.0 00 m. 00 03.0000 0
000.0000 000.0 00 50.00 —0 000.0 5 .00. F F ..0 000.000 000.0 00 m. 5 50.005 0
0.00. 5000 000.000 0.00.0000 000.000 50.000 0000.000 000.000 0. 5 000.0000 0
000.0000 000.000 000.00 ..0 600.000 000.000 N 5.00% 000.000 m. 00 0000.00 ..0
000.0000 000.000 000.0000 000.005 000.000 .000. .00 000.000 0.5 000.0000 0
000.0000 000.000 000.0000 000.000 000. 0000 000.000 000.000 0. ..0 000.0000 N
N00.0000 000.000 10.0.0000 000.000 000.000 000.000 000.000 0. 00 00.0000 0

 

 

 

Rama o... mmO

Ohm

>2.

mm

.>..8 000 5.0.0 000 3000200 000

140

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

wmodm Fm- mwod me 9&me mm F .NVNw whNdww mm F.Nm Fw mN
homdomw- nomdmmw gndme m FN®N Fm mmmJ. Fa Fv F .mmmm 6N
1\.NN..mKFw.. 6mm. FVVm Nmm. Foe medtm mwmde mvmdon mN
Nmothw- mmfimvnw Nmm. me nmqmme mmv.VNw mmdvmw NN
homfimw- mmodmvm mNo.Nvm MdeMNw modem MQOdN Fm FN
omnfih Fm- w 55wa momcwww medomm fimmdmw mmodon ON
fimmeNm- wmodoo. Fa VoFdow 2F. :3” NhF. Fvw mmo€ovw mF
momfiONm- com. 55% _va.mmw www.mmmw mthNw hmmdomw wF
wmmdomw- om F .Nmm$ FmNme mmvd me hmmfiNw MVNdmmw x. F
$060 Fm- 0 560mm .va. Fhw Fhv.mVNw 4&0in fiNan Fa o F
FmN. Fm Fa- 58%“me modmw F Fvfimw m 5.5% mondm Fa m F
ththm- mmwdmvw NmN. me Nmoémw onmdw wN F Nwa 6F
NmFdwmw- Nmmdomw 03.00% F Fmdmmw me.VNw wmm. 50% MF
mFm. Fth- 55.05.; thNww omndwmw medmw wmodFNw NF
mNFdNNw- €modmow www.mvw FoodmNm mm F. Fma m 36th FF
Nmn€m Fa- mm F 52% mmN. Kw 05.00wa wnmdmw wdem Fm o F
woodow- ENQNme o Fvfimw an F.N F Fm MNNde Fmvdma m
mnodhw- mmodmvw wmmfiww moN. FVNw Fo F 60 Fa mmmdmw w
munfimmm- gNdwv. Fa wadma F F twenm wVNde moo. F me h
NNVdm Fm- mmmfimvw mmNme Fmvfim Fa mNN. F Fw Nomdew m
mNm. Fm Fm- Nnodnvm {mucwom vméNNw 33.6% omNdn Fm m
mNN.mm Fm- ommdmmw N 5.0mm mndm Fm omwdmw mmm.v F Fm .v
mhvi FNw- wmmdmmw .33. Km womfmmw mnméo Fm 6mm. Fm Fa m
w Fm. Fme- who. Fume womdmw mmfimmmw mm F me w Fo.vaw N
vadmm- devmw @22wa 1 56wa OmNKme mom. FNNw F
3:03—3:50 3203:0230 mmo Ohm >2. mm

4535 0F

o._.

141

BIBLIOGRAPHY

. 1994. Activity-Based Costingfor Food Wholesalers and Retailers.
Joint Industry Project on Efficient Consumer Response, Ernst &
Young LLP, p. 4, 29.

. 2002. “ABC - Methodology and Practical Usage,” QPR Cost
Control, http://www.qpronline.com/abc/activity_based_intro.html.

Battezzati, Luigi, and Magnani, Roberto. 2000. “Supply Chains for FMCG
and Industrial Products in Italy,” International Journal of Physical
Distribution and Logistics Management, vol. 30, num. 5, p. 413-
424.

Bence, Don. 2001. "Choosing the Right Applications for Profitable Market
Share and Growth," Digital Printing for Packaging Conference
Proceedings, Packaging Digest, Converting, and PIRA International.

Bowersox, Donald J., and Closs, David J. 1996. Logistical Management:
The Integrated Supply Chain Process, McGraw-Hill Companies, Inc.

Bucklin, Louis P. 1965. “Postponement, Speculation, and the Structure
of Distribution Channels,” Journal of Marketing Research, vol. 2 num
2, p. 26-31.

Clarke, Robert H. 2001. "The Impact of Digital Printing on the Supply
Chain," Digital Printing for Packaging Conference Proceedings,
Packaging Digest, Converting, and PIRA International.

Closs, David J. 2003. Professor, Department of Marketing and Supply
Chain Management, Michigan State University, Series of Meetings.

Cohen, Jacob, Cohen, Patricia, West, Stephen G., and Aiken Leona S.
2003. Applied Multiple Regression/Correlation Analysis for the
Behavioral Sciences, 3rd Edition, Lawrence Erlbaum Associates,
Publishers, p. 479-480.

Cole, Nicholas. January 1986. "The Changing Shape of Distribution,"
Chief Executive, p. 28-32.

Dickeson, Roger V. 2000. “Change Your Business Model,” Printing
Impressions, vol. 43, issue 7, p. 118-119.

142

 

Emden, Reha. 2001—2003. Director, Digital Printing, The John Henry
Company, Series of Personal Interviews.

Falkman, Mary Ann. October 2001. "Digital Wins Favor with Packagers
and Converters," Packaging Digest, p. 81-84.

Feitzinger, Edward, and Lee, Hau L. 1997. "Mass Customization at
Hewlett-Packard: The Power of Postponement," Harvard Business
Review, vol. 75, num. 1, p. 116-121.

Fogarty, Donald W., Blackstone, John H. Jr., Hoffmann, Thomas R.
1991. Production and Inventory Management. South-Western
Publishing Co., p. 165-172.

Foster, Thomas A. 2000. “Eight Steps to a European Logistics Strategy,”
Logistics Management and Distribution Report, vol. 39, num. 4, p.
81-85.

Graman, Gregory A. 1999. “An Analysis of Capacitated Packaging
Postponement,” Ph.D. Dissertation, University of Cincinnati, p. 32-
33.

Hair, Joseph F. Jr., Anderson, Rolph E., Tatham, Ronald L., and Black,
William c. 1998. Multivariate Data AnalLsis, 5th Edition, Prentice
Hall, p.178.

Hills, William F., and Gugliotti, Jan. February 2001. "Reinventing
pharmaceutical research and development," Pharmaceutical
Executive, p. 14-17.

Howard, Kevin A. 1991. "Packaging Postponement Lowers Logistical
Costs," Logistical Packaging Innovation Symposium Proceedings,
Council of Logistics Management and School of Packaging at
Michigan State University, p. 123-131.

Johnson, M. Eric, and Anderson, Emily. 2000. “Postponement Strategies
for Channel Derivatives,” The International of Logistics
Management, vol. 11, num. 1, p. 19-36.

Stock, James R., and Lambert, Douglas M. 1982. Strategic Physical
Distribution Management. Irwin-McGraw Hill, p. 183.

143

Larson, Melissa. April 2000. “Ready or not, Here Comes Digital Printing,”
Converting Magazine, vol. 18, num. 4, p. 52-54.

Lee, Hau L., and Tang, Chistopher S. 1997. “Modeling the Costs and
Benefits of Delayed Product Differentiation,” Management Science,
vol. 43, num. 1, p. 40-53.

Lee, Hau L, Billington, Corey, and Carter, Brent. 1993. “Hewlett-Packard
Gains Control of Inventory and Service through Design for
Localization,” Interfaces, vol. 23, num. 4, p. 1-11.

Lewis, Howard T., Culliton, James W. and Steele, Jack D. 1956. “The
Role of Air Freight in Physical Distribution,” Boston: Division of
Research, Graduate School of Business Administration, Harvard
University.

McNeely, Susan B. 2001. "Digital Printing for Pharmaceutical Packaging -
A Customer Perspective," Digital Printing for Packaging Conference
Proceedings, Packaging Digest, Converting, and PIRA International.

Nahmias, Steven. 2001. Production and Operation Analysis, McGraw-
Hill/Irwin, p. 264.

Olsmats, Carl. 2000. “Future Demands in the Packaging Flow,”
Packaging Logistics Review. Edited by destrom, Kim, Econpap
Publications, vol. 1, p. 91-100.

Pagh, Janus D., and Cooper, Martha C. 1998. “Supply Chain
Postponement and Speculation Strategies: How to Choose the Right
Strategy,” Journal of Business Logistics, vol. 19, num. 2, p. 13-33.

Richardson, Helen L. 2000. “Packaging for Today’s Demanding
Fulfillment,” Transportation and Distribution. vol. 41, num. 7, p. 72-
74.

Richardson, Helen L. 2001. “Streamlining Fulfillment with Automation,”
Transportation and Distribution, vol. 42, num. 1, p. 28-30.

Schoenmakers, Wil. 2001. "How Are the Brand Owners Using Digital

Printing?" Digital Printing for Packaging Conference Proceedings,
Packaging Digest, Converting, and PIRA International.

144

Shapiro, Jeremy F. 1999. “Bottom-up vs. Top-down Approaches to

Supply Chain Modeling” Quantitative Models for Sgpply Chain
Management, Edited by Tayur, Ganeshan, and Magazine, Kluwer

Academic Publishers, p. 750-751.

Sox, Charles R., Thomas L. Joseph, and McClain, John O. 1997.
“Coordinating Production and Inventory to Improve Service,”
Management Science. vol. 43, num. 9, p. 1189-1197.

Trombley, Kurt. Senior Packaging Engineer, Procter & Gamble
Pharmaceuticals, Personal Interview, 2003.

Twede, Diana, and Parsons, Ben. 1997. “Distribution Packaging for
Logistical Systems,” PIRA International Reviews of Packagjm, p.
15-37.

Twede, Diana, Clarke, Robert H., and Tait, Jill A. 2000. “Packaging
Postponement: A Global Packaging Strategy,” Packaging Technology
and Science. vol. 13, p. 105-115.

Twede, Diana, and Selke, Susan E. M. 2003. Wood and Paper Packaging,
Working Textbook, Michigan State University.

Van Hoek, Remko I., Commandeur, Harry R., and Vos, Bart. 1998.
“Reconfiguring Logistics Systems Through Postponement
Strategies,” Journal of Business Logistics. vol. 19, num. 1, p. 33-54.

Van Hoek, Remko I. 2001. “The Rediscovery of Postponement a
Literature Review and Directions for Research," Journal of
Operations Management. vol. 19, num. 2, p. 161-184.

Waller, Matthew A., Dabholkar, Pratibha A., and Gentry, Julie J. 2000.
“Postponement, Product Customization, and Market-oriented Supply
Chain,” Journal of Business Logistics, vol. 21, num. 2, p. 133-160.

Wiersema, William H. 1995. Activity-Based Management. AMACOM, a
division of American Management Association, p. 7, 37, 39-48.

Yin, Robert K. 1994. Case Study Research DesiLn and Methods. Applied

Social Research Methods Series, vol. 5, 2nd Edition, Sage
Publications, p. 13.

145

 

Zinn, Walter, and Bowersox, Donald, J. 1988. "Planning Physical
Distribution with the Principle of Postponement," Journal of Business
Logistics, vol. 9, num. 2, p. 117-136.

Zinn, Walter. 1986. “An Analysis of the Cost and Benefit Associated with

the Principle of Postponement-Speculation,” Ph.D. Dissertation,
Michigan State University, p. 5-10, 57, 179-180, 262.

146