Dissertation Gerald Oeser

Methods of Risk Pooling in Business Logistics
and Their Application
Inauguraldissertation
zur Erlangung des akademischen Grades

„Doktor der Wirtschaftswissenschaften“
(Dr. rer. pol.)
eingereicht an der
Wirtschaftswissenschaftlichen Fakultät
der Europa-Universität Viadrina
in Frankfurt (Oder)
im September 2010
von
Gerald Oeser
Frankfurt (Oder)
“Uncertainty is the only certainty there is,
and knowing how to live with insecurity is the only security.”
John Allen Paulos,

Professor of Mathematics at Temple University in Philadelphia
To my brother,
who never failed to support me
Acknowledgements
First and foremost I would like to express my gratitude to my Ph.D. advisor Prof. Dr.
Dr. h.c. Knut Richter, Chair of General Business Administration, especially Industrial
Management, European University Viadrina Frankfurt (Oder), Germany for his extensive
mentoring. He also kindly gave me the opportunity of gaining experience in teaching grad-
uate courses in addition to my research.
For willingly delivering his expert opinion on my thesis as the second referee I thank
Prof. Dr. Joachim Käschel, Professorship of Production and Industrial Management, Tech-
nical University Chemnitz, Germany very much.
I would also like to gratefully acknowledge the support of all employees of Prof. Rich-
ter's chair, especially Dr. Grigory Pishchulov: Among other things, he helped me carry out
the “Subadditivity Proof for the Order Quantity Function (5.1)” in Appendix E mathemati-
cally formally correctly.
Furthermore, I thank Prof. Dr. Dr. h.c. Werner Delfmann, Director of the Department of
Business Policy and Logistics, University of Cologne, Germany for the opportunity offered
and for showing me that the world cannot be explained by mathematics alone.
Moreover, I appreciate the diverse support of the following professors during my Mas-
ter's and doctoral studies: Ravi M. Anupindi (University of Michigan), Ronald H. Ballou
(Case Western Reserve University), Gérard Cachon (University of Pennsylvania), Ian
Caddy (University of Western Sydney, Australia), Sunil Chopra (Northwestern Universi-
ty), Chandrasekhar Das (emeritus, University of Northern Iowa), David Dilts (Oregon
Health and Science University), Bruno Durand (University of Nantes, France), Philip T.
Evers (University of Maryland), Amit Eynan (University of Richmond), Dieter Feige
(emeritus, University of Nuremberg, Germany), Jaime Alonso Gomez (EGADE - Tec de
Monterrey, Mexico City and University of San Diego), Drummond Kahn (University of
Oregon), Peter Köchel (emeritus, Technical University Chemnitz, Germany), David H.
Maister (formerly Harvard Business School), Alan C. McKinnon (Heriot-Watt University,
UK), Esmail Mohebbi (University of West Florida), Steven Nahmias (Santa Clara Univer-
sity), Teofilo Ozuna, Jr. (The University of Texas-Pan American), David F. Pyke (Univer-
sity of San Diego), Pietro Romano (University of Udine, Italy), Wolfgang Schmid (Euro-
pean University Viadrina Frankfurt (Oder), Germany), Yossi Sheffi (Massachusetts Insti-
tute of Technology), Edward Silver (emeritus, University of Calgary, Canada), Jacky Yuk-
Chow So (University of Macau, China), Jayashankar M. Swaminathan (University of
North Carolina), Christian Terwiesch (University of Pennsylvania), Ulrich Thonemann
(University of Cologne, Germany), Biao Yang (University of York, UK), and Walter Zinn
(The Ohio State University).
I express gratitude to the paper merchant wholesaler Papierco for the insight into its op-
erations, the opportunity of optimizing its logistics, and the financial support.
I am also thankful to my friends: Without them writing this dissertation would have
been a lonesome experience.
Finally, I thank my father for constantly pushing and financially supporting me and
proofreading my thesis.
Geleitwort
Die vorgelegte Dissertation widmet sich dem großen Thema Risk Pooling in der Business-
Logistik. Die Aggregation von Risiken der Logistikprozesse ist ein derart universelles
Problem, so dass Autoren verschiedenster Wissenschafts- und Anwendungsgebiete dazu
ihre Positionen, Lösungsansätze und Modelle dargestellt haben. Bisher hatte noch niemand
den Mut aufgebracht, diese kaum überblickbare Menge an Literaturquellen zu sichten und
nach wissenschaftlichen Kriterien zu ordnen. Herr Oeser hat sich dieser Aufgabe gestellt
und hat sie mit Bravour gemeistert.
Bei aller Bedeutung dieser aufwendigen akademischen Aktivitäten darf jedoch nicht ver-
gessen werden, dass Risk Pooling ein Problem der Praxis ist, und dass die entsprechenden
theoretischen Untersuchungen letztendlich der Praxis dienen sollen. Die Vielfältigkeit des
Problems kann dabei kaum durch ein relevantes Basismodell erfasst werden. Es ist das
Verdienst des Autors, mit der Entwicklung einer Grundkonzeption für ein Entscheidungs-
unterstützungssystem zur Auswahl der geeigneten Risk Pooling Strategien Unternehmen
und besonders Unternehmensberatungen ein Instrument in die Hand zu geben, mit dem
sich die vielfältigen Situationen der Unternehmen analysieren lassen. Er belässt es jedoch
nicht beim Entwurf der Konzeption, sondern demonstriert auch überzeugend dessen An-
wendung anhand eines Unternehmens.
Es ließen sich viele weitere Vorzüge des hier veröffentlichten Werkes nennen. Es soll an
dieser Stelle jedoch dem Leser überlassen bleiben sich ein Urteil zu bilden. Herr Oeser
bietet dem Leser mit seiner Dissertation auf jeden Fall keine „leichte Kost“, denn der be-
handelte Gegenstand lässt eine oberflächliche Betrachtung nicht zu.
Wer sollte dieses Werk lesen? Wissenschaftler und Studenten aus den Gebieten Operations
Management, Logistik, Operations Research werden hier viel Neues und Interessantes,
auch viele Probleme für die weitere Forschung finden. Unternehmensberatungen werden
viele angeführte Lösungen zur Komplexitätsbewältigung in ihren Projekten anwenden
können. Ich bin überzeugt, dass die vorgelegte Dissertation neue Forschungen und Projekte
anregen wird. Das ist das Schönste, was sich ein Autor wohl wünschen kann.
Prof. Dr. Dr. h.c. Knut Richter

Table of Contents
List of Figures .................................................................................................................... X
List of Tables ..................................................................................................................... XI
List of Abbreviations, Signs, and Symbols ................................................................... XII
Abstract ................................................................................................................... XIV
1 Introduction ................................................................................................. 1
1.1 Problem: Growing Uncertainty in Business Logistics and

Need for a Tool to Choose Among Risk Pooling Countermeasures ............ 1
1.2 Objective ....................................................................................................... 3
1.3 Object and Method ........................................................................................ 3
1.4 Structure and Contents .................................................................................. 4
2 Risk Pooling in Business Logistics ............................................................. 6
2.1 Business Logistics Risk Pooling Research ................................................... 6

2.2 Placing Risk Pooling in the Supply Chain, Business Logistics,
and a Value Chain ......................................................................................... 8
2.3 Defining Risk Pooling ................................................................................. 11
2.4 Explaining Risk Pooling ............................................................................. 14
2.5 Characteristics of Risk Pooling ................................................................... 19
3 Methods of Risk Pooling ........................................................................... 24
3.1 Storage: Inventory Pooling ......................................................................... 27

3.2 Transportation ............................................................................................. 35
3.2.1 Virtual Pooling ............................................................................................ 35
3.2.2 Transshipments ........................................................................................... 35
3.3 Procurement ................................................................................................ 40
3.3.1 Centralized Ordering ................................................................................... 40
3.3.2 Order Splitting ............................................................................................ 41
3.4 Production ................................................................................................... 43
3.4.1 Component Commonality ........................................................................... 43
3.4.2 Postponement .............................................................................................. 45
3.4.3 Capacity Pooling ......................................................................................... 49
VII
Table of Contents
3.5 Sales and Distribution ................................................................................. 51

3.5.1 Product Pooling ........................................................................................... 51
3.5.2 Product Substitution .................................................................................... 52
4 Choosing Suitable Risk Pooling Methods ............................................... 53
4.1 Contingency Theory .................................................................................... 53

4.2 Conditions Favoring the Individual Risk Pooling Methods ....................... 54
4.3 The Risk Pooling Methods' Advantages, Disadvantages,
Performance, and Trade-Offs ...................................................................... 55
4.4 A Risk Pooling Decision Support Tool ...................................................... 56
5 Applying Risk Pooling at Papierco .......................................................... 74
5.1 Papierco ....................................................................................................... 74

5.2 Problems Papierco Faces ............................................................................ 75
5.2.1 Fierce Competition in German Paper Wholesale ........................................ 75
5.2.2 Supplier Lead Time Uncertainty ................................................................. 75
5.2.3 Customer Demand Uncertainty ................................................................... 76
5.2.4 Distribution Requirements Planning (DRP) ............................................... 80
5.2.5 Demand Forecasting Methods .................................................................... 82
5.3 Solving Papierco's Problems ....................................................................... 87
5.3.1 Determining Suitable Risk Pooling Methods for Papierco ......................... 87
5.3.2 Emergency Transshipments between Papierco's German Locations .......... 91
5.3.2.1 Optimizing Catchment Areas ...................................................................... 91
5.3.2.2 Increase in Transshipments and Its Causes ................................................. 94
5.3.2.3 Transshipments Are Worthwhile for Papierco ........................................... 96
5.3.3 Centralized Ordering ................................................................................. 101
5.3.3.1 Papierco's Current Order Policy ............................................................... 101
5.3.3.2 Stock-to-Demand Order Policy with Centralized Ordering and
Minimum Order and Saltus Quantities ..................................................... 105
5.3.3.3 Benefits of Centralized Ordering for Papierco ......................................... 107
5.3.4 Product Pooling ......................................................................................... 117
5.3.5 Inventory Pooling ...................................................................................... 118
5.3.6 Challenges in IT and Organization ........................................................... 121
5.4 Summary ................................................................................................... 123
6 Conclusion ............................................................................................... 126
VIII
Table of Contents
Appendix A: A Survey on Risk Pooling Knowledge and Application in

102 German Manufacturing and Trading Companies ........................ 134
1 Introduction ............................................................................................. 135
1.1 Motivation: Scarce Survey Research on Risk Pooling ............................. 135

1.2 Objective ................................................................................................... 140
2 The Survey ............................................................................................... 142
2.1 Research Design ........................................................................................ 142

2.2 Data Analysis and Findings ...................................................................... 147
2.2.1 Risk Pooling Knowledge and Utilization in
the German Sample Companies ................................................................ 147
2.2.2 Association between the Knowledge of
Different Risk Pooling Concepts .............................................................. 151
2.2.3 Association between Knowledge and Utilization of
Risk Pooling Concepts .............................................................................. 154
2.2.4 Association of the Utilization of Different Risk Pooling Concepts ........... 156
2.2.5 Risk Pooling Knowledge and Utilization in
the Responding Manufacturing and Trading Companies ......................... 161
2.2.6 Knowledge and Utilization of the Different Risk Pooling Concepts
in Small and Large Responding Companies ............................................. 164
3 Critical Appraisal .................................................................................... 166
4 Questionnaire .......................................................................................... 169
5 Answers .................................................................................................... 170
Appendix B: Proof of the Square Root Law for Regular, Safety, and
Total Stock ............................................................................................... 172
Appendix C: What Causes the Savings in Regular Stock through Centralization

Measured by the SRL? ........................................................................... 176
Appendix D: Tables ....................................................................................................... 178
Appendix E: Subadditivity Proof for the Order Quantity Function (5.1) ............... 202
Bibliography .................................................................................................................... 204
IX
List of Figures
Figure 2.1: Number of Publications on Risk Pooling in Business Logistics .................. 7

Figure 2.2: Placing Risk Pooling in Business Logistics .................................................. 9
Figure 2.3: Important Value Activities Using Risk Pooling Methods .......................... 11
Figure 4.1: Risk Pooling Decision Support Tool .......................................................... 56
Figure 5.1: Total Fine Paper Sales in Tons per Month in Germany
(BVdDP 2010a: 8) ...................................................................................... 76
Figure 5.2: Papierco's Total Fine Paper Sales to Customers in Germany
in Tons per Month ....................................................................................... 77
Figure 5.3: Papierco's 2007 Warehouse Sales to Customers per Week
of Seven Standard Paper Products from the Hemmingen Warehouse ........ 78
Figure 5.4: 2007 Incoming Goods, Warehouse Sales, and Inventory at
Warehouse Hemmingen .............................................................................. 80
Warehouse Reinbek .................................................................................... 81
Figure 5.6: Comparison of Current and Optimal Catchment Areas of Papierco's
Warehouses ................................................................................................. 93
Figure 5.7: Inventory Turnover Curves for Papierco .................................................. 104
Figure 5.8a: Product Purchase Planning at Papierco's Member Companies ................. 120
Figure 5.8b: Product Purchase Planning at Papierco's Member Companies ................. 120
Figure 5.9: Stockkeeping at Papierco's Warehouses in February 2008 ...................... 121
Figure A.1: Risk Pooling Knowledge and Utilization in
the German Sample Companies ................................................................ 147
Figure A.2: Risk Pooling Knowledge and Utilization in the
Sample Manufacturing and Trading Companies ...................................... 161
Figure A.3: Knowledge and Utilization of the Different Risk Pooling Concepts in
Small and Large Responding Companies ................................................. 164
Figure A.4: Questionnaire ............................................................................................ 169
X
List of Tables
Table 3.1: Risk Pooling Methods' Building Blocks ..................................................... 26

Table 5.1: The Extent of Transshipments at Papierco ................................................. 94
Table 5.2: Effects of Centralized Ordering at Papierco ............................................. 109
Table A.1: Comparing Respondents with the Total Population of
Manufacturing and Trading Companies in Germany ............................... 143
Table A.2: Significant Correlations at the 5 % Level of
the Knowledge of Risk Pooling Concepts ................................................ 154
Table A.3: Significant Correlations at the 5 % Level (except in the case Product
Substitution/Demand Reshape) between Knowledge and
Utilization of Risk Pooling Concepts ....................................................... 155
Table A.4: Significant Correlations at the 5 % Level between the Utilization of
Risk Pooling Concepts .............................................................................. 160
Table A.5: Survey Participants' Answers to the Questionnaire .................................. 171
Table D.1: Fulfillment of the SRL's Assumptions by Eleven Surveyed Companies .. 178
Table D.2: Comparison of Important Inventory Consolidation Effect,
Portfolio Effect, and Square Root Law Models ........................................ 179
Table D.3: Conditions Favoring the Various Risk Pooling Methods ......................... 185
Table D.4 The Risk Pooling Methods' Advantages, Disadvantages,
Performance, and Trade-Offs .................................................................... 195
XI
List of Abbreviations, Signs, and Symbols
Common general, business, logistics, supply chain management (SCM), operations re-
search (OR), mathematical, and statistical abbreviations, signs, and symbols apply1 and are
not listed here.
CC component commonality
CE consolidation effect
CO centralized ordering
COE centralized ordering effect
COV coefficient of variation of demand
CP capacity pooling
CS cycle stock
Dipl.-Kfm. Diplomkaufmann (MBA equivalent)
DP demand pooling
DUR demand uncertainty reduction
EFR effective fill rate for the customer
ELT emergency lateral transshipment
EOS economies of scale
FR item fill rate
IC inventory holding cost
i. i. d. independent and identically distributed
IP inventory pooling
ITO in terms of
LP lead time pooling
LT lead time
LTUR lead time or lead time uncertainty reduction
n. e. c. not elsewhere classified
NLT endogenous lead times
OP order policy
OS order splitting
1
Cf. e. g. Soanes and Hawker (2008), Friedman (2007), Lowe (2002), Obal (2006), Silver et al. (1998) and
Slack (1999), Clapham and Nicholson (2009), and Dodge (2006).
XII
List of Abbreviations, Signs, and Symbols
PCE portfolio cost effect

PE portfolio effect
PM postponement
PoD point of (product) differentiation
PP product pooling
PQE portfolio quantity effect
PRE proportional reduction of error
PS product substitution
RMSE root mean square error
RPDST risk pooling decision support tool
RS regular stock
SL transshipment sending location
SLA service level adjustment
sqrt square root
SRL square root law
SS safety stock
TC transportation cost
TS transshipments
TSC transshipment cost
VP virtual pooling
vs. versus, here: is traded off against
XLT exogenous lead times
≈ is approximately equal to
≻ is preferred to
XIII
Abstract
Purpose/topicality: Demand and lead time uncertainty in business logistics increase, but can
be mitigated by risk pooling. Risk pooling can reduce costs for a given service level, which is
especially valuable in the current economic downturn. The extensive, but fragmented and in-
consistent risk pooling literature has grown particularly in the last years. It mostly deals with
specific mathematical models and does not compare the various risk pooling methods in terms
of their suitability for specific conditions.
Approach: Therefore this treatise provides an integrated review of research on risk
pooling, notably on inventory pooling and the square root law, according to a value-chain
structure. It identifies ten major risk pooling methods and develops tools to compare and
choose between them for different economic conditions following a contingency approach.
These tools are applied to a German paper merchant wholesaler, which suffers from cus-
tomer demand and supplier lead time uncertainty. Finally, a survey explores the knowledge
and usage of the various risk pooling concepts and their associations in 102 German manu-
facturing and trading companies. Triangulation (combining literature, example, modeling,
and survey research) enhances our investigation.
Originality/value: For the first time this research presents (1) a comprehensive and
concise definition of risk pooling distinguishing between variability, uncertainty, and risk,
(2) a classification, characterization, and juxtaposition of risk pooling methods in business
logistics on the basis of value activities and their uncertainty reduction abilities, (3) a deci-
sion support tool to choose between risk pooling methods based on a contingency ap-
proach, (4) an application of risk pooling methods at a German paper wholesaler, and (5) a
survey on the knowledge and utilization of risk pooling concepts and their associations in
102 German manufacturing and trading companies.
XIV
1 Introduction
1.1 Problem: Growing Uncertainty in Business Logistics and

Need for a Tool to Choose Among Risk Pooling
Countermeasures
Product variety has increased dramatically in almost every industry2 particularly due to in-
creased customization3. Product life cycles have become shorter, demand fluctuations more
rapid, and products can be found and compared easily on the internet.4
This causes difficulties in forecasting for an increased number of products, demand and
lead time uncertainty, intensified pressure for product availability, and higher inventory
levels5 to provide the same service6. This trend is expected to continue and likely grow
worse.7
Supply chains8 are more susceptible to disturbances today because of their globaliza-
tion, increased dependence on outsourcing and partnerships, single sourcing, little leeway
in the supply chain, and increasing global competition.9 Disruptions, such as production or
shipment delays, affect profitability (growth in operating income, sales, costs, assets, and
inventory).10
Risk pooling is “[o]ne of the most powerful tools used to address [demand and/or lead
time] variability in the supply chain”11 particularly in a period of economic downturn12, as
it allows to reduce costs and to increase competitiveness.
2
Cox and Alm (1998), Van Hoek (1998a: 95), Aviv und Federgruen (2001a), Ihde (2001: 36), Piontek
(2007: 86), Ganesh et al. (2008: 1124f.).
3
Ihde (2001: 36), Chopra and Meindl (2007: 305), Piontek (2007: 86).
4
Rabinovich and Evers (2003a: 226), Chopra and Meindl (2007: 305, 333).
5
Dubelaar et al. (2001: 96).
6
Ihde (2001: 36), Swaminathan (2001: 125ff.), Chopra and Meindl (2007: 305, 333), Rumyantsev and
Netessine (2007: 1)
7
Cecere and Keltz (2008).
8
“A supply chain consists of all parties involved, directly or indirectly, in fulfilling a customer request.
The supply chain includes not only the manufacturer and suppliers, but also transporters, warehouses, re-
tailers, and even customers themselves” (Chopra and Meindl 2007: 3).
9
Dilts (2005: 21). These ideas of Professor David M. Dilts, Owen Graduate School of Management, Van-
derbilt University, have not appeared in a formal publication yet. He granted us permission to cite them as
personal correspondence on July 23, 2008.
10
Hendricks and Singhal (2002, 2003, 2005a, 2005b, 2005c), Hoffman (2005).
11
Simchi-Levi et al. (2008: 48). We will differentiate between the often confused terms variability, uncer-
tainty, and risk in section 2.3.
12
Cf. Hoffman (2008), Chain Drug Review (2009a), Hamstra (2009), Orgel (2009), Pinto (2009b), Ryan
(2009).
1
1 Introduction
Although “risk pooling is often central to many recent operational innovations and
strategies”13, an important concept in business logistics and supply chain management
(SCM)14, and was already described in logistics in 196715, it is mentioned in few German
text books16 mostly limited to the square root law (SRL). Most other publications also only
consider inventory pooling17 or a single other type18 of risk pooling19. Exceptions are Neale
et al. (2003: 44f., 50, 55), Fleischmann et al. (2004: 70, 93), Taylor (2004: 301ff.), Muck-
stadt (2005: 150ff.), Reiner (2005: 434), Anupindi et al. (2006), Heil (2006), Brandimarte
and Zotteri (2007: 30, 36, 38, 39, 57, 58, 70), Sheffi (2007), Simchi-Levi et al. (2008), So-
bel (2008: 172), Van Mieghem (2008), Cachon and Terwiesch (2009), and Bidgoli (2010).
Our survey of 102 German manufacturing and trading companies of various sizes and
industries (see appendix A20) showed that the different risk pooling concepts are known
fairly well, but not widely applied despite their potential benefits. Choosing risk pooling
methods is difficult for companies, as the literature does not compare the various methods
in terms of their suitability for certain conditions holistically.21 Most of the work in risk
13
Cachon and Terwiesch (2009: 350).
14
Romano (2006: 320).
15
Flaks (1967: 266).
16
For example Pfohl (2004a: 115ff.), Tempelmeier (2006: 41f., 153f.), Bretzke (2008: 147, 152, 154).
17
For example Bramel and Simchi-Levi (1997: 219f.), Martinez et al. (2002: 14), Christopher and Peck
(2003: 132), Dekker et al. (2004: 32), Ghiani et al. (2004: 9), Daskin et al. (2005: 53f.), Li (2007: 210),
Taylor (2008: 13-3), Mathaisel (2009: 19), Shah (2009: 89f.), Wisner et al. (2009: 513).
18
A type is “a category of […] things having common characteristics” (Soanes and Hawker 2008). We
distinguish ten main types of risk pooling that have in common that they may reduce total demand and/or
lead time variability and thus uncertainty and risk by pooling individual demand and/or lead time varia-
bilities. We call them methods (e. g. in the title of this treatise), whenever we focus on the ways of risk
pooling, as a method is “a way of doing something” (Soanes and Hawker 2008). If the focus is on choos-
ing and implementing one or several risk pooling methods for a specific company under specific condi-
tions, we refer to this as risk pooling strategy. A strategy is “a plan designed to achieve a particular long-
term aim” (Soanes and Hawker 2008). In the case of risk pooling this plan can comprise one or more risk
pooling methods combined in order to reduce demand and/or lead time uncertainty. We use the term con-
cept, if we aim at the “abstract idea” (Soanes and Hawker 2008) or would like to include the SRL, PE,
and inventory turnover curve, which are rather tools to measure the risk pooling effect on inventories than
risk pooling methods.
19
Pfohl (2004b: 126, 355), Enarsson (2006: 178).
20
For the sake of readability and content unity we placed the survey in the appendix.
21
However, Evers (1999) compares inventory centralization, transshipments, and order splitting. Swamina-
than (2001) designs a framework for deciding between part and procurement standardization (component
commonality), process standardization (postponement), and product standardization (substitution) accord-
ing to product and process modularity, Wanke and Saliby (2009: 690) one for choosing between invento-
ry centralization (demand pooling) and regular transshipments (“serving […] demands from all centra-
lized facilities” and enabling demand and lead time pooling). Benjaafar et al. (2004a, 2005) find in sys-
tems with endogenous supply lead times, multisourcing (Benjaafar et al. 2004a: 1441f., 1446) and capaci-
ty pooling (Benjaafar et al. 2005: 550, 563) perform better than inventory pooling, if utilization (arrival
rate divided by service rate) is high. Eynan and Fouque (2005) show that demand reshape is more effi-
cient than component commonality.
2
1 Introduction
pooling is rather focused on one aspect and holistic treatments are rare.22 There is little
empirical and – to our knowledge – no survey research on the various methods of risk
pooling combined. Most publications develop mathematical models for a specific risk
pooling method under certain assumptions and (optimal inventory) policies that minimize
inventory for a given service level or maximize service level for a given inventory. They
do not explore whether this risk pooling method is the best for the given situation or
whether it can be combined with other ones.
1.2 Objective
Thus the objective of this treatise is to advance research on and aid practice in selecting and
applying risk pooling methods in business logistics. More specifically, the aims of this research
can be formulated as follows:
1. Develop the first comprehensive and concise definition of risk pooling.
2. Critically review and structure the research on risk pooling within a value-chain
framework according to this definition.
3. Develop tools to compare and choose appropriate risk pooling methods and mod-
els for different economic conditions with a contingency approach.
4. Apply these tools to a German paper merchant wholesaler, which faces customer
demand and supplier lead time uncertainty.
5. Examine the knowledge and usage of the various risk pooling concepts and their
associations in 102 German trading and manufacturing companies by means of a
survey.
1.3 Object and Method
The research object (German: Erfahrungsgegenstand) is an empirical phenomenon (a part of

reality) that is to be described. The scientific object or selection principle (German: Er-
kenntnisgegenstand) is the perspective and specific question used to examine the research ob-
ject.23
Our research object comprises business logistics in a company that experiences de-
mand and/or lead time uncertainty. Our scientific object is risk pooling that can decrease
this uncertainty of the company, which is perceived as a collection of value activities (val-
22
Personal correspondence with Professor Christian Terwiesch, The Wharton School, University of Penn-
sylvania on July 2, 2008.
23
Schneider (1987: 34f.), Chmielewicz (1994: 19ff.), Söllner (2008: 5), Neus (2009: 2).
3
1 Introduction
ue chain approach). Business logistics plans, organizes, handles, and controls all material,
product, and information flows across these value activities24 in an efficient, effective, and
customer-oriented manner (logistics perspective). Our main specific research questions
are: What risk pooling methods are available? What economic conditions make the indi-
vidual methods worthwhile? How can a company choose adequate risk pooling methods?
We derive the adequacy of the identified risk pooling methods from contingency factors
(contingency approach).
In order to enhance the quality of our research we use triangulation25 to gather various
types of information26 on risk pooling by literature, example, modeling, and survey re-
search. This integrated approach also accounts for our holistic ambition and leads to results
not attainable by the individual research methods separately27. Weaknesses of some me-
thods can be balanced by the strengths of others.28 Van Hoek (2001: 182f.) already pro-
posed triangulation to improve research on postponement, one type of risk pooling. For
problems of triangulation, such as coping with conflicting results, epistemological prob-
lems, and the debatable higher validity of findings, please refer to e. g. Bryman (1992:
63f.), Denzin and Lincoln (2005: 912), Blaikie (1991: 122f.), and Fielding and Fielding
(1986: 33). A holistic approach may not be as profound as an atomistic one. However, as
noted before such a holistic approach is novel to risk pooling research.
1.4 Structure and Contents
This treatise is divided into six chapters and an appendix. After this introduction (chapter 1),
we outline previous risk pooling research in business logistics, identify ten main types of risk
pooling, and place them in the supply chain, business logistics, and a value chain. This results
in the five logistics value activities storage, transportation, procurement, production, and sales
and distribution, where risk pooling is important. Completing outlining our research frame-
work, we define, explain, and characterize risk pooling in business logistics (chapter 2).
Chapter 3 defines, classifies, and characterizes the ten identified risk pooling methods
within the five logistics value activities and according to their uncertainty reduction abilities.
24
To Porter (2008: 75) inbound and outbound logistics are primary value activities themselves besides op-
erations, marketing and sales, and service. We adopt a cross-functional view of logistics.
25
Denzin (1978: 291) defines triangulation as “the combination of methodologies in the study of the same
phenomenon”.
26
Graman and Magazine (2006: 1070).
27
Jick (1979: 603f.), Fielding and Fielding (1986: 33), Graman and Magazine (2006: 1070).
28
Blaikie (1991: 115).
4
1 Introduction
Research on inventory pooling is reviewed in more detail29, as it is the earliest and most
extensive and prominent one. In particular, we attempt to remedy inconsistencies regarding
the SRL, a tool to estimate the inventory savings from risk pooling by inventory pooling:
Under what assumptions can the SRL be applied to regular, safety, and total stock, as re-
searchers do not agree on them? What causes the inventory savings measured by the SRL?
Are its results confirmed in practice? Is the SRL questioned justifiedly? After examining the
portfolio effect (PE), a generalization of the SRL, and the inventory turnover curve, we com-
pare the various SRL and PE models in a synopsis (table D.2) based on their assumptions
and assign them to groups. This facilitates choosing an appropriate model to determine stock
savings from inventory pooling for specific conditions or adapt the existing models by add-
ing or dropping assumptions.
Chapter 4 merges the results of our literature review in a synopsis about conditions
making the various risk pooling methods favorable, their advantages, disadvantages, and
basic trade-offs. Based on this synopsis and contingency theory, a Risk Pooling Decision
Support Tool is developed for choosing an appropriate risk pooling strategy for a specific
company and specific conditions.
This tool is applied to a German paper wholesaler in chapter 5 to choose suitable risk
pooling methods to cope with demand and lead time uncertainty. In this context we also
show that in contrast to Maister (1976: 132) and Evers (1995: 2, 14f.) centralization or cen-
tralized ordering can reduce cycle stocks, if the replacement principle or stock-to-demand
replenishment policy is followed in case minimum order and saltus quantities have to be
observed (section 5.3.3.2).
If risk pooling is shown to be beneficial in theory and in practice also for this German
paper wholesaler, to what extent are its different methods known, applied, and associated in
102 German manufacturing and trading companies? To answer this question we present a
survey in appendix A for the above reasons. Such a survey has not been conducted before
and is demanded by some authors30.
The conclusion (chapter 6) summarizes our findings and identifies avenues for further re-
search.
29
Professor Walter Zinn, Fisher College of Business, The Ohio State University, believes this is sorely
needed (personal correspondence on July 4, 2008).
30
For example by Thomas and Tyworth (2006: 254f.) for order splitting and by Huang and Li (2008b: 19)
for postponement.
5
2 Risk Pooling in Business Logistics
This chapter outlines our research framework: It gives an overview of previous risk pooling
research (section 2.1), identifies ten main types of risk pooling and embeds them in the supply
chain, business logistics, and a value chain (section 2.2), which sets the structure for chapter 3,
defines risk pooling in business logistics (section 2.3), explains its mathematical and statistical
foundations (section 2.4) and examines five general features of risk pooling (section 2.5).
2.1 Business Logistics Risk Pooling Research
Risk pooling in business logistics is an active field of research with well over 600 publica-
tions so far. The number of publications in this area has steadily increased since its scarce be-
ginnings in the 1960s, gained momentum especially in the 80s and 90s and reached its peak in
2003 as figure 2.1 shows. Perhaps driven by the economic downturn, research activity in 2009
was the highest after 2003. Most research focused on inventory pooling (centralization, the
square root law, portfolio effect, and inventory turnover curve), postponement and delayed
(product) differentiation, and transshipments and inventory sharing. Figure 2.1 is based on the
literature databases Business Source Complete, Business Source Premier, EconLit, and Re-
gional Business News accessed via EBSCOhost® January 11, 2010.
After the concept borrowed from modern portfolio and insurance theory has been main-
ly applied to inventory centralization, meanwhile research focuses on other forms of risk
pooling, especially postponement and transshipments, and the coordination of risk pool-
ing arrangements and cost and profit allocation through contracts31 and fair allocation
rules, schemes, or methods32 with game theory.
The academics dealing with risk pooling in business logistics are of different back-
grounds (mathematics, operations research (OR), operations management (OM), man-
agement science, decision sciences, statistics, computer sciences, engineering, business
administration, management, economics, game theory, production, marketing, logis-
tics/supply chain management, physical distribution, and transportation), orientation
(quantitative, analytical, modeling, simulation, empirical, and qualitative research), and
31
Dana and Spier (2001), Cheng et al. (2002), Cachon (2003), Wang et al. (2004), Bartholdi and
Kemahloğlu-Ziya (2005), Cachon and Lariviere (2005), Özen et al. (2008, 2010).
32
Gerchak and Gupta (1991), Robinson (1993), Hartman and Dror (1996), Anupindi et al. (2001), Lehrer
(2002), Granot and Sošić (2003), Ben-Zvi and Gerchak (2007), Dror and Hartman (2007), Wong et al.
(2007a), Dror et al. (2008). Robinson (1993), Raghunathan (2003), Kemahloğlu-Ziya (2004), Bartholdi
and Kemahloğlu-Ziya (2005), and Sošić (2006) consider allocations based on the Shapley (1953) value.
6
prominence. Therefore they use inconsistent terms33, frameworks, and structures and
made our thorough literature review and definitions necessary. One might argue that due
to these differences this research must not be compared and only major research should be
cited. However, in contrast to previous research we would like to pursue a more holistic
approach to convey an integrated overview of business logistics risk pooling. Even less
prominent research can draw attention to important details and show directions for further
research. Some often cited risk pooling research has not been published in highly ranked
journals.34 Nevertheless most of the cited references are from the latter.
Most risk pooling research is quantitative and designs mathematical models of prob-
lems, develops solution methods (exact methods, algorithms, simulation methods, and heu-
ristics), and determines solutions (optimal inventory control and risk pooling, e. g. trans-
shipment, policies).
60
Inventory Pooling (150)
Postponement (105)
Transshipments (95)
50 Component Commonality (55)
Virtual Pooling (51)
Capacity Pooling (41)
Product Substitution (40)
40
Number of Publications
Order Splitting (37)

Centralized Ordering (27)
Product Pooling (21)
30
20
10
0
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Year of Publication
Figure 2.1: Number of Publications on Risk Pooling in Business Logistics
33
Bender et al. (2004: 233) made similar observations for location theory.
34
For example Eppen and Schrage (1981).
7
2.2 Placing Risk Pooling in the Supply Chain,

Business Logistics, and a Value Chain
Business logistics35 comprises the holistic, market-conform, and efficient planning, organiza-
tion, handling, and control of all material, product, and information flows from the supplier to
the company, within the company, and from the company to the customer36 and back (reverse
logistics37).
As figure 2.2 shows, the efficient and effective material, product, and information flow
is impaired by inter alia demand and lead time uncertainty. Risk pooling methods can miti-
gate these uncertainties. They can be implemented at or between the various supply chain
members (suppliers, manufacturers, wholesalers, distribution centers, central warehouses,
delivery or regional warehouses, retailers, stores, and customers).
Thoroughly reviewing the literature referred to in section 2.1, we identified ten risk
pooling methods: capacity pooling, central ordering, component commonality, inven-
tory pooling, order splitting, postponement, product pooling, product substitution,
transshipments, and virtual pooling. We will define and characterize them in detail in
chapter 3.
They can be implemented everywhere along the supply chain. Component commonality
rather concerns production. In general capacity and inventory pooling and postponement
may pool inventories or capacities of or for different locations. Component commonality,
postponement, product pooling, and product substitution may refer to products or their
components. Transshipments and virtual pooling deal with product, material, and informa-
tion flows between supply chain members within an echelon or across echelons, and cen-
tral ordering and order splitting with procurement between supply chain members.
In our opinion, all mentioned risk pooling methods but order splitting can reduce de-
mand uncertainty, order splitting only lead time uncertainty, component commonality,
capacity pooling, inventory pooling, product pooling, and product substitution only de-
35
Business logistics (management) is difficult to distinguish from supply chain management (SCM). The
terms are often used synonymously, although logistics is majorly seen as a part of SCM and not vice ver-
sa as in earlier literature (Ballou 2004b: 4, 6f., Larson and Halldórsson 2004, CSCMP 2010). According
to Kotzab (2000) the German business logistics conception already comprised a holistic management
along the whole value creation chain before it adopted the English term SCM. Gabler's business enzyclo-
pedia defines SCM as building and administrating integrated logistics chains (material and information
flows) from raw materials production via processing to end consumers (Alisch et al. 2004: 2870).
36
Wegner (1996: 8f.).
37
See e. g. Richter (1996a, 1996b, 1997), Richter and Dobos (1999, 2003a, 2003b), Dobos and Richter
(2000), Richter and Sombrutzki (2000), Richter and Weber (2001), Richter and Gobsch (2005), Gobsch
(2007).
8
mand uncertainty. Transshipments and virtual pooling, postponement, and central ordering
may dampen both uncertainties. Concerning order splitting and component commonality
there are dissenting individual opinions, which we will elaborate on in section 4.4.
Various economic conditions have been found to favor the different risk pooling me-
thods. We will focus on these favorable conditions in chapter 4. They flow into a Risk
Pooling Decision Support Tool that helps to determine suitable risk pooling methods for
specific conditions (section 4.4). This tool is applied to determine adequate risk pooling
methods for a German paper merchant wholesaler for coping with its demand and lead
time uncertainty in chapter 5.
Demand Uncertainty
Manufac-
Suppliers Distributors Retailers Customers
turers
Lead Time Uncertainty
Economic
Conditions
Material, product, and information flow
Risk pooling methods
Figure 2.2: Placing Risk Pooling in Business Logistics

Supply chain member names based on Chopra and Meindl (2007: 5).
Logistics may be considered as a cross-organizational (figure 2.2) and at every member of the
above supply chain a cross-departmental coordination function across all divisions, especially
storage, transportation, procurement, production of goods and services (including R&D,
recycling, and remanufacturing), and sales and distribution (including order processing, re-
covery, return, and disposal)38 in the value chain in figure 2.3.
38
Delfmann (2000: 323f.), Ballou (2004b: 9ff., 27, 29), Kuhlang and Matyas (2005: 659f.), Grün et al.
(2009: 15, 305), Wannenwetsch (2009: 21f.).
9
The value chain is a management concept that was developed by Porter (1985: 33ff.)
and describes a company as a collection of activities. These activities create value, use re-
sources, and are linked in processes. In our value chain, the main value activities are pro-
curement, production, and sales and distribution, which are supported by the value activi-
ties transportation and storage. Value activities are “technologically and economically dis-
tinct activities [… a company] performs to do business”39.
Inventory pooling (IP) mainly pertains to storage, virtual pooling (VP) via drop-
shipping and cross-filling and transshipments (TS) to transportation, centralized ordering
(CO) and order splitting (OS) to procurement, capacity pooling (CP), component commo-
nality (CC), and (form) postponement (PM) to production, and product pooling (PP) and
product substitution (PS) to sales. However, VP, extending a location's inventories to other
locations' ones, is also related to storage and sales, transportation and sales or service ca-
pacities may be pooled, any logistics decision may be postponed, and PP and PS may be
applied in production as well. This not mutually exclusive and exhaustive classification is
reflected in figure 2.3 and the next chapter's structure. We structured our review according
to the value chain approach, because we focus on the impact of risk pooling on the five
mentioned logistics value activities.
Risk pooling helps a company to cope with demand and/or lead time uncertainty and
thus to carry out these value activities at a lower cost for a given service level, a higher
service level for a given cost, or a combination of both40. Thus it may increase expected
profit41 by reducing expected costs and/or increasing expected revenue.
It may allow a company to win a competitive advantage over its competitors by effec-
tively combining Porter's (2008: 75) competition strategies of differentiation and cost lea-
dership, e. g. in mass customization enabled by postponement42.
39
Porter (2008: 75)
40
Cf. Chopra and Meindl (2007: 336), Cachon and Terwiesch (2009: 325, 350).
41
Porter's (1985: 38) value chain considers margin instead of profit. “Margin is the difference between total
value and the collective cost of performing the value activities”.
42
For example Feitzinger and Lee (1997).
10
Transportation
CP, PM, TS, VP
Sales &
Production
Procurement Distribution
CC, CP, PM, PP, Profit
CO, OS, PM CP, PM, PP,
PS
PS, VP
Storage
IP, PM, VP
Figure 2.3: Important Value Activities Using Risk Pooling Methods
2.3 Defining Risk Pooling
The literature offers various definitions of and confusion about the terms variability, variance,
or volatility43, uncertainty44, and risk45. Lead time and demand uncertainty may arise from
lead time and demand variability or incomplete knowledge46. “Uncertainty is the inability to
determine the true state of affairs of a system”47. “Uncertainty caused by variability is a result
of inherent fluctuations or differences in the quantity of concern. More precisely, variability
occurs when the quantity of concern is not a specific value but rather a population of values”48.
Lead time and demand uncertainty may lead to economic risk49, the possibility50 of a negative
deviation from expected values or desired targets51. The corporate target is expected profit
(figure 2.3), the difference of expected revenue and expected cost.52 The possibility of a posi-
tive deviation from an expected value constitutes a chance.53
Despite the costs risk pooling entails54, it may reduce variability and thus uncertainty
and expected (ordering, inventory holding, stockout, and backorder) costs55 and/or increase
expected revenue (product availability, fill rate, service level)56 and thus expected profit57.
43
Hubbard (2009: 84f.). “Outside of finance, volatility may not necessarily entail risk—this excludes consi-
dering volatility alone as synonymous with risk” (Hubbard 2009: 91).
44
Knight (2005: 19ff.), Haimes (2009: 265ff.), Hubbard (2010: 49f.).
45
Wagner (1997: 51), Knight (2005: 19ff.), Hubbard (2009: 79ff., 2010: 49f.).
46
Cf. Haimes (2009: 265). For a detailed treatment of the confusion about the terms uncertainty and varia-
bility please refer to Haimes (2009: 265ff.).
47
Haimes (2009: 265).
48
Haimes (2009: 266).
49
Bowersox et al. (1986: 58), Delfmann (1999: 195), Pishchulov (2008: 17).
50
Wagner (1997: 51).
51
Cf. e. g. Wagner (1997: 51), Köhne (2007: 321).
52
Wagner (1997: 52).
53
Wagner (1997: 51).
54
Kim and Benjaafar (2002: 16), Cachon and Terwiesch (2009: 328).
11
We define risk pooling in business logistics as consolidating individual variabilities

(measured with the standard deviation58) of demand and/or lead time in order to re-
duce the total variability59 they form and thus uncertainty and risk60 (the possibility
of not achieving business objectives61). The individual variabilities are consolidated by
aggregating62 demands63 (demand pooling64) and/or lead times65 (lead time pooling66).
Consolidating and aggregating mean “combining several different elements […] into a
whole”67.
As individual variabilities68 and not individual risks69 are pooled, the term risk pooling
is misleading. Nonetheless, we use it, because it is conventional.
55
Eppen (1979), Chen and Lin (1989), Tagaras (1989), Tagaras and Cohen (1992: 1080f.), Evers (1996:
114, 1997: 71f.), Cherikh (2000: 755), Eynan and Fouque (2003: 704), Kemahloğlu-Ziya (2004), Bar-
tholdi and Kemahloğlu-Ziya (2005), Wong (2005), Jiang et al. (2006: 25), Thomas and Tyworth (2006:
253), Chopra and Meindl (2007: 324ff.), Pishchulov (2008: 8, 17f.), Schmitt et al. (2008a: 14, 20), Sim-
chi-Levi et al. (2008: 48), Cachon and Terwiesch (2009: 325, 331, 344, 350).
56
Krishnan and Rao (1965), Tagaras (1989), Tagaras and Cohen (1992: 1080f.), Evers (1996: 111, 114,
1997: 71f., 1999: 122), Eynan (1999), Cherikh (2000: 755), Ballou and Burnetas (2000, 2003), Xu et al.
(2003), Ballou (2004b: 385-389), Wong (2005), Kroll (2006), Reyes and Meade (2006), Chopra and
Meindl (2007: 324ff.), Cachon and Terwiesch (2009: 325, 329, 344, 350).
57
Anupindi and Bassok (1999), Cherikh (2000: 755), Lin et al. (2001a), Eynan and Fouque (2003: 704, 707,
2005: 98), Kemahloğlu-Ziya (2004), Bartholdi and Kemahloğlu-Ziya (2005), Özen et al. (2005), Chopra
and Meindl (2007: 324ff.), Simchi-Levi et al. (2008: 234-238), Cachon and Terwiesch (2009: 331), Yang
and Schrage (2009: 837).
58
Sussams (1986: 8), Romano (2006: 320). “The standard deviation is the most commonly used and the
most important measure of variability” (Gravetter and Wallnau 2008: 109). Zinn et al. (1989: 2) and Cho-
pra and Meindl (2007: 307) consider the standard deviation a measure of uncertainty. Cachon and Ter-
wiesch (2009: 331f., 282f.) and Chopra and Meindl (2007: 307) use the derived coefficient of variation
(standard deviation divided by mean) as a measure for demand variability or uncertainty.
Pooling independent random variables does not change total variability measured with the variance. Of
course, one could argue that the measuring unit of the variance is squared and therefore difficult to interp-
ret and that the standard deviation and not the variance is used to calculate safety stock. Pooling variabili-
ties measured with the range may even increase total variability.
59
Cf. e. g. Chopra and Meindl (2007: 336).
60
Cf. e. g. Pishchulov (2008: 18).
61
Wagner (1997: 51).
62
Cf. e. g. Anupindi et al. (2006: 167), Chopra and Meindl (2007: 336).
63
Gerchak and Mossman (1992: 804), Swaminathan (2001: 131), Hillier (2002b: 570), Randall et al. (2002:
56), Gerchak and He (2003: 1027), Özer (2003: 269), Chopra and Sodhi (2004: 55, 60), Nahmias (2005:
334), Anupindi et al. (2006: 167, 187), Çömez et al. (2006), Romano (2006: 320), Chopra and Meindl
(2007: 177), Pishchulov (2008: 8, 18, 26), Simchi-Levi et al. (2008: 48, 196, 281, 348), Yu et al.
(2008: 1), Yang and Schrage (2009: 837), Bidgoli (2010: 209).
64
Evers (1997: 55, 57, 1999: 121f.), Benjaafar and Kim (2001), Benjaafar et al. (2004a: 1442, 2004b: 91),
Chopra and Sodhi (2004: 59ff.), Tomlin and Wang (2005: 37), Gürbüz et al. (2007: 302), Van Mieghem
(2007: 1270f.), Ganesh et al. (2008: 1134), Cachon and Terwiesch (2009: 332), Wanke and Saliby (2009:
690), Yang and Schrage (2009: 837).
65
Thomas and Tyworth (2006: 254).
66
Evers (1999: 121f.), Cachon and Terwiesch (2009: 336).
67
Soanes and Hawker (2008).
68
Gerchak and He (2003: 1028).
69
The business logistics risk pooling literature finds “the risk related to the uncertainty” of individual de-
mands (Zinn 1990: 12), risk “over uncertainty in customer demand” (Anupindi and Bassok 1999: 187),
standard deviations (Zinn 1990: 13) or variances of individual demands (Zinn 1990: 16), risk “over de-
mand uncertainty” (Weng 1999: 75) or risk “over random supply lead time” (Weng 1999: 82), “inventory
12
Among others Hempel (1970: 654), Wacker (2004: 630), and the references they give
make requirements for a “good”70 definition. To our knowledge previous attempts to de-
fine risk pooling do not satisfy them. They merely describe its causes71, effects72, or aim73,
only target demand pooling74, and equate risk pooling with inventory pooling75 and the
square root law76. Moreover they do neither define nor differentiate between variability77,
uncertainty, and risk. The business logistics risk pooling literature states risk pooling re-
duces
 variability78, “lead-time variability”79, lead time demand variability80, demand va-
riability81, demand variation82, variation83, “demand variance”84, variance of deli-
very time85, “the variance of the retailers' net inventory processes”86, “the mean and
variance of cycle stock”87,
 uncertainty88, demand uncertainty89, “the uncertainty the firm faces”90, “the effect
of uncertainty”91,
risks” (Van Hoek 2001: 174, Pishchulov 2008: 27), “inventory risk” (Kemahlioğlu-Ziya 2004: 76), “fore-
cast risk” (Chopra and Sodhi 2004: 58, 60), risk (Schwarz 1989: 830, 832, McGavin et al. 1993: 1093,
Chopra and Sodhi 2004: 58, 60f., Simchi-Levi et al. 2008: 357, Yang and Schrage 2009: 837), risks (Aviv
and Federgruen 2001a: 514, Gerchak and He 2003: 1027), “risk over the outside-supplier leadtime”
(Schwarz 1989: 828, McGavin et al. 1993: 1093), “lead-time risk” (Thomas and Tyworth 2006: 245f.,
2007: 169), “lead-time uncertainty” (Evers 1997: 70, Thomas and Tyworth 2006: 245), “lead-time fluctu-
ations” (Evers 1997: 70), “demand uncertainty” (Evers 1997: 70), uncertainty (Evers 1994: 51, Rabino-
vich and Evers 2003a: 206), or “quantity and timing uncertainty” (Collier 1982: 1303) is or are pooled.
70
“A ‘good’ [formal conceptual] definition […] is a concise, clear verbal expression of a unique concept
that can be used for strict empirical testing” (Wacker 2004: 631). Hempel (1970: 654) requires inclusivi-
ty, exclusivity, differentiability, clarity, communicability, consistency, and parsimony.
71
Nahmias (2005: 334), Anupindi et al. (2006: 168), Romano (2006: 320), Simchi-Levi et al. (2008: 48),
Wisner et al. (2009: 513), Bidgoli (2010: 209).
72
Gerchak and He (2003: 1027), Özer (2003: 269), Romano (2006: 320), Simchi-Levi et al. (2008: 48),
Cachon and Terwiesch (2009: 325, 350), Bidgoli (2010: 209).
73
Chopra and Meindl (2007: 212), Cachon and Terwiesch (2009: 321).
74
Flaks (1967: 266), Gerchak and Mossman (1992: 804), Gerchak and He (2003: 1027), Özer (2003: 269),
Nahmias (2005: 334), Anupindi et al. (2006: 187), Chopra and Meindl (2007: 212).
75
Anupindi et al. (2006: 168).
76
Wisner et al. (2009: 513).
77
Tallon (1993: 192, 199f.) equates variability with uncertainty.
78
Eynan and Fouque (2005: 91), Anupindi et al. (2006: 167).
79
80
Benjaafar and Kim (2001).
81
Flaks (1967: 266), Zinn (1990: 11), Randall et al. (2002: 56), Eynan and Fouque (2003: 705, 2005: 91),
Romano (2006: 320), Thomas and Tyworth (2006: 255, 2007: 188), Chopra and Meindl (2007: 212),
Pishchulov (2008: 18), Simchi-Levi et al. (2008: 48), Cachon and Terwiesch (2009: 325, 331), Wisner et
al. (2009: 513), Bidgoli (2010: 209).
82
Eynan and Fouque (2005: 91), Nahmias (2005: 334).
83
Evers (1999: 133).
84
Kemahlioğlu-Ziya (2004: 71).
85
Masters (1980: 71), Ihde (2001: 33).
86
Schwarz (1989: 830).
87
Thomas and Tyworth (2006: 247f.)
88
Pishchulov (2008: 22), Simchi-Levi et al. (2008: 281).
89
Eynan and Fouque (2003: 714), Pishchulov (2008: 17).
13
 “the risk associated to the variability”92, the “impact of individual risks”93, “risks
associated with forecasting errors and inventory mismanagement”94, and “inventory
risk”95.
Risk pooling is described “to hedge uncertainty so that the firm is in a better position to miti-
gate the consequence of uncertainty”96, “enables one to avoid […] uncertainty”97, or “removes
some of the uncertainty involved in planning stock levels”98.
It is also referred to as “statistical economies of scale”99, “portfolio efficiencies”100,
“Pooling Efficiency trough Aggregation” or “Principle of Aggregation”101, and “IMPACT
OF AGGREGATION ON SAFETY INVENTORY”102.
Risk pooling is also considered a form of operational hedging. “Hedging is the action
of a decision maker to mitigate a particular risk exposure. Operational hedging is risk miti-
gation using operational instruments”103, e. g. pure diversification or demand pooling104.
For more on operational hedging please refer to Boyabatli and Toktay (2004) and Van
Mieghem (2008: 313-350).
2.4 Explaining Risk Pooling
Risk pooling can be shown e. g. for inventory or location pooling: Let a single product be
stocked at n separate locations. Demand for this product is a normal random variable105 xi with
known mean µi and standard deviation106 σi for each location i = 1, …, n. The standard devia-
90
91
Özer (2003: 269).
92
Risk pooling “is applied to portfolio theory in finance here [sic!] the risk associated to the variability in
the return from individual stocks is diluted when an investor keeps a portfolio of stocks“ (Zinn 1990: 12).
93
Dilts (2005: 23).
94
Yang and Schrage (2009: 837).
95
Chopra and Sodhi (2004: 59).
96
97
Pishchulov (2008: 26) remarks this for risk pooling through delayed differentiation.
98
Jackson and Muckstadt (1989: 2).
99
Eppen (1979: 498), Eppen and Schrage (1981: 52), Evers (1994: 51), Özer (2003: 269), Rabinovich and
Evers (2003a: 206).
100
Eppen and Schrage (1981: 52).
101
Anupindi et al. (2006: 187, 189).
102
Chopra and Meindl (2007: 318).
103
Van Mieghem (2007: 1270).
104
Van Mieghem (2007: 1270f.).
105
A random variable is a variable that takes its values (realizations) with certain probabilities respectively
whose values are assigned to certain probability densities (Alisch et al. 2004: 3454).
106
If (the empirical distribution of) demand is forecast (Thonemann 2005: 255f.), σi is the standard deviation
of the distribution of the forecast error in formula (2.1) for calculating safety stock (Caron and Marchet
1996: 239, Pfohl 2004a: 114, Thonemann 2005: 255f., Chopra and Meindl 2007: 306). An estimate of
expected demand (forecast value) is ordered to satisfy the expected value of demand and safety stock is
14
tion σi is a measure of dispersion of individual values of the random variable xi around the
mean µi for every entity i and therefore a measure of xi's variability107.
If each location just satisfies its own demand, location i needs to hold an amount of
safety stock that allows it to hedge against the demand uncertainty associated with xi. Let
the optimal safety stock at location i in accordance with the newsboy model108 be
, (2.1)
where z is the safety factor that corresponds to a certain target service level. Therefore,
the total safety stock across all locations is
∑ . (2.2)
If all inventory holding is centralized at one location, this location needs to serve the to-
tal demand
∑ . (2.3)
The individual demands are aggregated across all locations. Safety stock in the centra-
lized system is
, (2.4)
where
109
∑ 2∑ ∑ (2.5)
is the standard deviation of x and ρij is the correlation coefficient of the value of the random
variable for locations i and j. It can be formally shown that the aggregated variability (standard
deviation of total demand σa) is less than or equal to the sum of the individual variabilities
built up as protection against the forecast error, which is at least as high as the demand uncertainty or
standard deviation of demand (personal correspondence with Professor Ulrich W. Thonemann, University
of Cologne, in 2008), and not against uncertainty in demand (Thonemann 2005: 255f.). A high safety
stock is needed, if the (standard deviation of the) forecast error is high. The size of demand fluctuations is
irrelevant (Thonemann 2005: 257). If the distribution of demand is known, σi is the standard deviation of
demand (Zinn et al. 1989: 4) and safety stock is held to hedge against uncertainty in demand. The higher
the uncertainty in demand, the higher is the safety stock. The standard deviation of demand is zero and no
safety stock is needed, if there is no demand uncertainty (Thonemann 2005: 238) and no lead time uncer-
tainty either. Nonetheless, some companies forecast demand, but wrongly use the standard deviation of
demand instead of the standard deviation of the forecast error in calculating safety stock (Korovessi and
Linninger 2006: 489f.).
107
Gravetter and Wallnau (2008: 109).
108
See e. g. Thonemann (2005: 220), Cachon and Terwiesch (2009: 235).
109
Cf. Mood et al. (1974: 178), Zinn et al. (1989: 5), Jorion (2009: 43). This expression is also written
∑ 2∑ ∑ (Eppen 1979: 500) or ∑ 2∑ ∑ (cf.
Weisstein 2010). “The double summation sign ∑ ∑ indicates that all possible combinations of i and j
should be included in calculating the total value” (Moyer et al. 1992: 222), where j is larger than i.
15
(sum of standard deviations of demand at the n locations σ) because of the subadditivity prop-
erty of the square root of non-negative real numbers110:
∑ ∑ 2∑ ∑ . (2.6)
Therefore the safety stock in the centralized system is less than or equal to the one in the
decentralized one
∑ ∑ 2∑ ∑ . (2.7)
Inequality (2.6) is a special case of the known Minkowski inequality for p = 2. It is al-
ways correct, if the variances exist, therefore also for the Poisson and Binomial distribu-
tion.111
Hence, the standard deviation of the aggregate demand is lower than or equal to the sum
of the standard deviations of the individual demands. Consequently, inventory pooling or
centralization at a single location can reduce the amount of safety stock necessary to en-
sure a given service level. The reduction in safety stock depends on the correlations be-
tween xi, i = 1, …, n. Inventory pooling does not always reduce safety stock due to the
less-than-or-equal sign.
Yet, the sum of the individual variabilities (standard deviations) only equals the total
aggregated variability (the square root of the sum of the individual variances plus two
times the covariance of the random variable's value for two entities i and j) in two cases:
(1) The random variables xi are perfectly positively correlated (the coefficient of
correlation ρij equals 1, i, j):
∑ 2∑ ∑ ∑ ∑ 112
(2.8)
∑ ∑ 1 ∑ ∑ .
(2) Random variables xi cannot mutually balance their fluctuations, if at least

n-1 σi equal zero: If n-1 σi equal zero, (2.6) becomes
(2.9)
for this single non-zero σi. If all σi are zero, (2.6) becomes
110
Gaukler (2007).
111
Minkowski (1896), Abramowitz and Stegun (1972: 11), Bauer (1974: 72), Gradshteyn and Ryzhik (2007:
1061).
112
Cf. Moyer et al. (1992: 222). This expression is also written ∑ ∑ ∑ , (cf.
Jorion 2009: 43).
16
0 . (2.10)
Apart from this, for independent (the correlation coefficient ρij is equal to 0, i, j) nor-
mally distributed random variables risk pooling leads to variability reduction:
∑ ∑ . (2.11)
The highest possible variability reduction is achieved, if there are negative correla-
tions which make the second term under the square root equal to minus the first term113 in
(2.6).
Some authors114 give the impression that risk pooling always reduces total variability or
enables to reduce inventory, although this must not be the case as shown above.
Likewise industry and academia often assume that inventory pooling, a type of risk
pooling, always is beneficial, i. e. it either reduces costs or increases profits, and that the
value of inventory pooling increases with increasing variability of demand.115
Kemahlioğlu-Ziya (2004: 40) states this was only always correct for normally distributed
demand such as in Eppen (1979) or Eppen and Schrage (1981). She neglects though that
this is not correct for perfectly positively correlated demands and if at least n-1 σi equal
zero.
Furthermore, for uncertain demand and certain conditions more willingness to substitute
may not lead to higher expected profits or “lower optimal total inventory”116 for full117 and
partial substitution or risk pooling118.
113
Cf. Eppen (1979: 500).
114
Tallon (1993: 186) imprecisely remarks, “Mathematically, the square root of the sum of the variances is
less than the sum of the individual standard deviations”. Likewise, Anupindi et al. (2006: 191) inaccu-
rately state “the inventory benefits” from physical centralization “result from the statistical principle
called the principle of aggregation, which states that the standard deviation of the sum of random vari-
ables is less than the sum of the individual standard deviations”. Gaukler (2007) uses the less-than-or-
equal sign, but inconsistently says the standard deviation of the aggregate demand was lower than (not
lower than or equal to) the sum of the standard deviations of the individual demands. Although Gaukler
(2007) states his remarks were not intended to be a rigorous derivation of this concept, they should be
consistent. “It is well known that the pooled demand has a lower standard deviation than the sum of stan-
dard deviations of individual demands. Thus, the safety stock as well as inventory holding and shortage
costs are lower when products are more substitutable” (Ganesh et al. 2008: 1134). “Inventory pooling
represents a strategy of consolidating inventories and aggregating stochastic demands, enabling reduc-
tions in inventory holding and shortage costs” (Pishchulov 2008: 8). “Risk pooling suggests that demand
variability is reduced if one aggregates demand across locations” (Romano 2006: 320, Simchi-Levi et al.
2008: 48). “The aggregation of demand stemming from risk pooling leads to reduction in demand varia-
bility, and thus a decrease in safety stock and average inventory” (Simchi-Levi et al. 2008: 50). “Centra-
lizing inventory reduces both safety stock and average inventory in the system” (Simchi-Levi et al. 2008:
51). Finally, inventory pooling does not automatically reduce inventory, but it may allow to reduce the
inventory necessary to provide a given service level.
115
Kemahlioğlu-Ziya (2004: 40).
116
117
Baker et al. (1986), Pasternack and Drezner (1991), Gerchak and Mossman (1992).
118
McGillivary and Silver (1978), Parlar and Goyal (1984), Anupindi and Bassok (1999), Ernst and Kouve-
lis (1999), Rajaram and Tang (2001), Netessine and Rudi (2003).
17
Yang and Schrage (2009) show this “inventory anomaly” (after demand substitution or
risk pooling the company ought to hold more stock rather than less) for full substitution
with any right skewed demand distribution119 if the shortage is somewhat higher than the
holding cost per unit, for partial substitution with exponential, Normal, Poisson, and uni-
form demands even if they are negatively correlated, as well as for more than one period,
with backlogging, lost sales, more than two products, and with setup costs. The inventory
anomaly is increasing as the underage is close to the overage cost per unit and as the cor-
relation between demands decreases.120 The relative difference of the shortage to holding
cost for which the anomaly arises increases with increasing right skewedness. A company
will not face the inventory anomaly, if it uses a shortage penalty, but minimizes inventory
carrying costs subject to a fixed target service level. However, it may not seize the chance
to increase sales and profits, if it augments demand pooling without raising inventory. The
likelihood of the inventory anomaly is higher with short lead times and lower with less
frequent but larger orders, e. g. with high setup costs.121
Inventory pooling may reduce costs and increase profits for the supply chain party hold-
ing inventory122, but may reduce the total supply chain profits. In a two-echelon supply
chain, where the upper echelon (the supplier) carries inventory, the lower echelon (the re-
tailers), whose revenues depend only on sales, may lose profits due to pooling.123 The sup-
plier and the retailers are likely to benefit from risk pooling inventory, if the stockout pe-
nalty cost is high.124
The normal random variable xi in our derivation of risk pooling can also be the demands
for i unique components, product versions, substitute products, customized products, or
(replenishment) lead times to i locations. They are aggregated to the demand for a common
component in component commonality125, universal product in product pooling, all substi-
tutes in product substitution or demand reshape, the undifferentiated generic product in
postponement or delayed product differentiation, or to an aggregated lead time across all
locations, suppliers, or deliveries in lead time pooling (emergency transshipments and or-
der splitting). The aggregated demand and/or lead time x may fluctuate less, as the stochas-
119
“Actual demand distributions tend to be right skewed” (Yang and Schrage 2009: 847).
120
121
122
Kemahloğlu-Ziya (2004), Bartholdi and Kemahloğlu-Ziya (2005).
123
Anupindi and Bassok (1999).
124
Dai et al. (2008: 411).
125
Dogramaci (1979: 130) shows that “the standard deviation of demand for the common component (σc)
would be less than or equal to” the sum of “the standard deviation[s] of lead time demand” of the compo-
nents it replaces.
18
tic fluctuations (σi) of the individual demands and/or lead times usually balance each other
to a certain extent (equation 2.6).
Risk pooling by demand pooling in transshipments, virtual pooling, centralized order-
ing, and capacity pooling can be derived in the same manner as shown for inventory pool-
ing: Demands are pooled across locations.
2.5 Characteristics of Risk Pooling
We now turn to describing five important characteristics common to most risk pooling me-
thods: (1) increasing returns, but (2) diminishing marginal returns with increasing application
as well as increasing benefit with (3) increasing demand variability, (4) decreasing demand
correlation, and (5) decreasing concentration of uncertainty.
(1) The benefit of or return on risk pooling is variability reduction and thus enabled in-
ventory reduction for a given service level or increase in service level for a given invento-
ry. The risk pooling return generally (in the following cases) increases with increasing
application or the number of participants:
It augments with the number of
 participating stock-keeping locations in inventory pooling126, time and logistics
postponement127, transshipments128, and cross-filling129,
 (substitutable) products and degree of substitution in product substitution, demand
reshape130, and resource flexibility131,
 products in the product line132 or products being postponed133 in manufacturing
postponement,
 products sharing components (increasing commonality) in component commonali-
ty134, as well as
126
Lin et al. (2001a), Heil (2006: 170), Netessine and Rudi (2006), Milner and Kouvelis (2007), Cachon and
Terwiesch (2009: 282f.).
127
Zinn and Bowersox (1988: 133), Heil (2006: 170).
128
Jönsson and Silver (1987a: 224), Robinson (1990), Diks and de Kok (1996: 378), Tagaras (1999), Evers
(2001: 313), Herer et al. (2002), Tagaras and Vlachos (2002), Dong and Rudi (2004), Burton and Baner-
jee (2005), Hu et al. (2005), Heil (2006: 170).
129
Ballou and Burnetas (2000, 2003), Ballou (2004b: 385-389).
130
Eynan and Fouque (2003, 2005), Ganesh et al. (2008: 1124).
131
Tomlin and Wang (2005: 51).
132
Zinn (1990: 14), Swaminathan and Tayur (1998).
133
134
Collier (1982: 1303), Srinivasan et al. (1992: 26), Grotzinger et al. (1993: 532f.), Eynan and Fouque
(2005), Chew et al. (2006: 247).
19
 retailers in pooling over the outside-supplier lead time or centralized ordering135

and risk pooling by drop-shipping or virtual pooling136.
(2) However, the marginal benefit of risk pooling commonly decreases with each addi-
tional increase in risk pooling or participant. There appear to be diminishing marginal
returns to risk pooling137, e. g. to cross-filling138 and transshipments with increasing num-
ber of locations transshipping139 or increasingly even allocation of demand across loca-
tions140, to virtual pooling by drop-shipping with increasing number of retailers141, to com-
ponent commonality142, postponement143, product substitution144, capacity pooling, and
location pooling145.
The marginal benefit of location pooling decreases with increasing number of pooled
locations146, so that the main benefit is gained by consolidating a few locations and it might
not be necessary to pool all locations147. The same applies to transshipments.148
Capacity pooling with a little bit of flexibility149 as long as it is designed with long
chains150 almost has the same benefit as full flexibility. Companies can benefit from any
135
Jackson and Muckstadt (1989: 14), Kumar and Schwarz (1995), Gürbüz et al. (2007).
136
Netessine and Rudi (2006: 845).
137
Similarly, in corporate finance we can observe diminishing marginal risk reduction with increasing num-
ber of securities: At first diversification reduces portfolio risk (standard deviation or volatility) rapidly,
then more slowly with increasing number of randomly chosen stocks (Statman 1987: 353, 355). In recent
years stocks have become individually more volatile but are increasingly less than perfectly correlated.
Therefore an investor needs to hold more stocks to capture the majority of benefits from diversification
than before (Campbell et al. 2001: 40).
138
Ballou and Burnetas (2000, 2003), Ballou (2004b: 385-389). Based on Ballou's (2003b: 81) numerical
example, we devise the general formula for the effective fill rate EFR for the customer in dependence of
the for all locations equal item fill rate FR (0 1) and the number n of locations that cross-fill as
1 1 . The first derivative of EFR with respect to n shows that the more locations cross-
fill the higher is the EFR: ⁄ 1 · ln 1 0. However, the incremental in-
crease in EFR diminishes with increasing number of locations taking part in cross-filling: ⁄
1 · ln 1 · ln 1 0.
139
Evers (1997: 70).
140
Evers (1996: 128).
141
142
Chopra and Meindl (2007: 327f.) find by a numerical example that “the marginal benefit [… marginal
reduction in safety stock] decreases with increasing commonality” for independent and normally distri-
buted end-product demand, but formulate this result as if generally valid. Bagchi and Gutierrez (1992)
show by two numerical examples (817) that “increasing component commonality results in increasing
marginal returns when the criteria are aggregate service level and aggregate stock requirement” (815) for
two end-products that face independent, identically exponentially, geometrically, or Poisson distributed
demands and share up to three components (815, 817f., 824). They do not give an intuitive explanation
(829).
143
Zinn (1990: 14), Graman and Magazine (2006: 1075, 1080).
144
Eynan and Fouque (2003: 707, 2005: 96).
145
Cachon and Terwiesch (2009: 323, 349).
146
147
Cachon and Terwiesch (2009: 323f.).
148
Tagaras (1989), Evers (1997: 71f.), Ballou and Burnetas (2000, 2003), Ballou (2004b: 385-389), Kranen-
burg and Van Houtum (2009).
20
amount of risk pooling as long as they implement it appropriately151 and demand is not
perfectly positively correlated. Graman and Magazine (2006: 1075) similarly observe per-
taining to postponement “that it is only necessary to postpone a portion of a product to
realize most of the benefits of such a strategy”. “The mathematical inventory model shows
that almost all of the positive benefits of postponement (such as lower inventories) can be
achieved with a partial (low-capacity) postponement scenario”152.
Cachon and Terwiesch (2009: 324) show diminishing marginal returns of location pool-
ing with increasing mean for independent Poisson demands.153 This must not necessarily
hold for other distributions such as the normal one.154
(3) The (demand) risk pooling effect decreases with increasing demand correlation155
and any Magnitude (relative sizes of the standard deviations of demand), and increases
149
Flexibility means that a plant is capable of producing more than one product. With no flexibility each
plant can only produce one product; with total flexibility every plant can produce every product (as in
manufacturing postponement). Graphically lines called links indicate which plant is capable of producing
which product. Flexibility allows production shifts to high selling products to avoid lost sales (Cachon
and Terwiesch 2009: 344f.).
150
“A chain is a group of plants and [… products] that are connected via links” (Cachon and Terwiesch
2009: 346).
151
152
153
“[T]he standard deviation of a Poisson distribution equals the square root of its mean. Therefore, Coeffi-
√
cient of variation of a Poisson distribution […].
√
As the mean of a Poisson distribution increases, its coefficient of variation decreases, that is, the Poisson
distribution becomes less variable. Less variable demand leads to less inventory for any given service
level. Hence, combining locations with Poisson demand reduces the required inventory investment be-
cause a higher demand rate implies less variable demand. However, because the coefficient of variation
decreases with the square root of the mean, it decreases at a decreasing rate. In other words, each incre-
mental increase in the mean has a proportionally smaller impact on the coefficient of variation, and hence
on the expected inventory investment” (Cachon and Terwiesch 2006: 324). This can also be derived for-
mally: The first derivative of the coefficient of variation ⁄ √ ⁄ with respect to the mean λ
is smaller than zero, the second one larger than zero: ⁄ 0, ⁄ 0.
154
For the normal distribution it depends on how the coefficient of variation changes with respect to the
mean. With the Poisson distribution it decreases as the mean increases, but this is not necessarily the case
with the normal distribution. However, we find that there are at least diminishing marginal returns to
pooling of normally distributed independent demands with the same standard deviation and mean. For
normally distributed independent demand the coefficient of variation of pooled demand is
∑
∑
. If the standard deviation and mean is the same at each pooled location i, this expression simpli-
√
fies to . The first and second derivative of COV with respect to the number n of
√
pooled locations shows that as n increases COV decreases at a decreasing rate: ⁄ 0,
⁄ 0.
155
Eppen (1979), Zinn et al. (1989: 12), Inderfurth (1991: 109, 111), Netessine and Rudi (2003: 329), Alfaro
and Corbett (2003), Corbett and Rajaram (2004, 2006), Heil (2006: 170), Chopra and Meindl (2007: 319),
Cachon and Terwiesch (2009: 349). In serial or divergent multi-stage base-stock production/distribution
systems the lower the demand correlation, the further upstream safety stocks should be held and vice ver-
sa and the lower the worth of safety stocks because of product-dependent risk diversification (risk pool-
21
with decreasing correlation and Magnitude156. Likewise the value of lead time risk pooling
(order splitting157 and transshipments) increases with decreasing correlation of replenish-
ment lead times.
Therefore, many “companies attempt to reduce inventory and manufacturing costs by
manipulating correlated demand sources, or inducing demand patterns that balance each
other in an average sense”158, i. e. by risk pooling, while maintaining sales.159
The bullwhip effect160 can be reduced by risk pooling (effects)161, production smooth-
ing, and seasonality162, so that it may be overestimated163.
Evers and Beier (1993) and Evers (1995) debatably show that there are no further sav-
ings in safety stock after a first consolidation because of the arising perfectly positive cor-
relation.
Tyagi and Das (1999: 211) show that if demands are correlated and one appropriately
takes advantage of their characteristics, a partial may be more cost-efficient than complete
pooling of customers.
On the other hand, Xu and Evers (2003) claim that partial can never outperform com-
plete pooling only based on demand correlation. Examples suggesting that one should pre-
fer partial to complete aggregation were based on inconsistent correlation matrices.
(4) The benefits of risk pooling generally increase with a structured (“a common linear
transformation”164) increase in demand variability165, if lead times are exogenous (for in-
ventory pooling, product consolidation, and delayed differentiation166). Gerchak and He
ing) (Inderfurth 1991: 109, 111). Nonetheless, the total safety stock size may increase with decreasing
correlation due to the impact of time-dependent risk diversification on carrying safety stocks at the still
more costly final stage level because of long replenishment lead times (Inderfurth 1991: 111).
156
Zinn et al. (1989: 3, 12), cf. Tallon (1993: 201), Tyagi and Das (1998: 201), Wanke (2009: 122).
157
Minner (2003: 273), Thomas and Tyworth (2006: 254, 2007: 188).
158
Burer and Dror (2006: 1).
159
Gerchak and Gupta (1991), Ruusunen et al. (1991), Hartman and Dror (2003: 243f.), Tyagi and Das
(1999).
160
The bullwhip effect describes that demand (order) variabilities are amplified as they move upstream the
supply chain (Lee et al. 1997a: 93, 1997b: 546, 2004: 1875, Sucky 2009: 311). It is caused by demand
forecast updating or demand signaling, order batching, price fluctuation, as well as rationing and shortage
gaming (Lee et al. 1997a: 95, 1997b: 546, 2004: 1883) and can be counteracted by avoiding multiple de-
mand forecast updates, breaking order batches, stabilizing prices, and eliminating gaming in shortage sit-
uations (Lee et al. 1997a: 98ff., 1997b: 555ff., 2004: 1883ff.).
161
Hartman and Dror (2003: 244), Lee et al. (2004: 1882), Cachon et al. (2007: 475), Chen and Lee (2009:
781), Sucky (2009: 311).
162
Cachon et al. (2007).
163
Cachon et al. (2007), Sucky (2009).
164
165
Kemahlioğlu-Ziya (2004: 22), Zinn et al. (1989), Alfaro and Corbett (2003), Eppen (1979).
166
Benjaafar and Kim (2001: 19f.), Benjaafar et al. (2004b: 90f., 2005).
22
(2003) provide a newsvendor counterexample with convex ordering where increased va-
riability of two individual demands reduces the benefits of risk pooling.
If supply lead times are endogenous in multi-item finite capacity production-inventory
systems, both higher demand variability167 and capacity utilization (arrival rate divided by
service rate)168 or asymmetric backordering or holding costs make risk (inventory) pool-
ing169 less valuable170.
(5) Assuming a multivariate normal distribution and uncorrelated demands, the value of
pooling is lower when uncertainty is more concentrated, i. e. the less uniform or more
dispersed the values of the standard deviations of the demands are.171 Correspondingly,
“the PE will be larger when variances are of similar rather than highly varying magni-
tude”172.
167
Benjaafar and Kim (2001: 19), Kim and Benjaafar (2002: 15), Benjaafar et al. (2004a: 1446).
168
Benjaafar and Kim (2001: 19), Kim and Benjaafar (2002: 15), Benjaafar et al. (2004a: 1438, 2004b: 71,
2005: 550).
169
“While the relative benefit of inventory pooling tends to diminish with utilization, the relative benefit of
capacity pooling tends to increase with utilization” (Benjaafar et al. 2005: 550).
170
Kim and Benjaafar (2002: 15).
171
Alfaro and Corbett (2003: 24).
172
Tyagi and Das (1998: 201), cf. Zinn et al. (1989).
23
3 Methods of Risk Pooling
After clarifying the prerequisites in the previous chapter, we will now enumerate the ten identi-
fied risk pooling methods and their synonyms, present a classification according to their ability
to pool demands and/or lead times, and define, classify, and characterize them correspondingly
within the structure of the aforementioned five logistics value activities.
First, under the heading storage we deal with inventory pooling and centralization, the
SRL, PE, and inventory turnover curve, as this area shows the earliest and most extensive
research in risk pooling (cf. figure 2.1) and requires a review173. The various SRL and PE
models are compared in a synopsis (table D.2) based on their assumptions. This facilitates
choosing an appropriate model to determine stock savings from inventory pooling for spe-
cific conditions or adapt the existing models by adding or dropping assumptions and re-
lates our centralized ordering considerations in section 5.3.3.2 to these models.
Afterwards, virtual pooling, which is related to inventory pooling, and transshipments
are examined (transportation). Upon considering risk pooling in these two original (tradi-
tional) logistics duties that refer to the transfer of objects (above all of goods)174, we survey
it in the derivative logistics areas175 procurement (centralized ordering and order splitting),
production (component commonality, postponement, and capacity pooling), and sales and
distribution (product pooling and substitution) downstream the value chain.
Perhaps since risk pooling and centralization of inventories (distribution or warehouse
networks) are closely related, the terms inventory pooling176 or just pooling177 are often
used interchangeably and ambiguously for risk pooling (methods), although inventory
pooling literally merely refers to the consolidation or centralization of inventory, is a
type178 or manifestation179 of risk pooling, and takes advantage of the possible benefits of
risk pooling.
173
Personal correspondence with Professor Walter Zinn, Fisher College of Business, The Ohio State Univer-
sity on July 4, 2008.
174
Delfmann (2000: 323).
175
Delfmann (2000: 323).
176
Benjaafar and Kim (2001: 13), Kemahlioğlu-Ziya (2004: 11), Gaur and Ravindran (2006: 482), Pishchu-
lov (2008: iii, 23f., 26, 28ff., 133), Simchi-Levi et al. (2008: 239).
177
Evers (1999: 121), Benjaafar and Kim (2001: 13), Alfaro and Corbett (2003: 12), Kemahlioğlu-Ziya
(2004: 11), Pishchulov (2008: iv, vi, vii, 8, 17f.).
178
“A finite set of outlets with randomly fluctuating demands bands together to reduce costs by buying,
storing and distributing their inventory jointly. This is termed inventory centralization and is a type of risk
pooling” (Burer and Dror 2006: 1). Cachon and Terwiesch (2009: 321).
179
24
After a thorough literature review we conclude that apart from

(1) inventory pooling180, location pooling181, or (physical) inventory or ware-
house centralization182 (SRL, PE, selective stock keeping, inventory turnover
curve),
risk pooling in business logistics can also be achieved by
(2) virtual pooling183, virtual inventory or inventories184, “electronic invento-
ry”185, virtual centralization186, virtual warehouse187 or warehousing188, “vir-
tually aggregating inventories”189, or information centralization190,
(3) transshipments (stock sharing between inventory locations)191 or “location
substitution”192,
(4) centralized ordering193, “consolidated distribution”194, “risk pooling over the
outside-supplier lead time”195, or ”warehouse risk-pooling”196,
(5) order splitting or multiple suppliers197,
(6) component commonality198, component sharing199, part standardization200, or
procurement standardization201,
180
Benjaafar and Kim (2001: 13), Alfaro and Corbett (2003: 13), Anupindi et al. (2006: 189, 273), Heil
(2006), Yang and Schrage (2009: 837). Cf. also references in section 1.1.
181
Cachon and Terwiesch (2009: 321ff.).
182
Eppen (1979), Zinn (1990: 12), Benjaafar and Kim (2001: 13), Martinez et al. (2002: 14), Ghiani et al.
(2004: 9), Reiner (2005: 434), Anupindi et al. (2006: 187-194), Sheffi (2006: 117f., 2007), Brandimarte
and Zotteri (2007: 57f.), Chopra and Meindl (2007: 336), Pishchulov (2008: iii, 17, 133).
183
Cachon and Terwiesch (2009: 321, 350, 469), Mahar et al. (2009: 561).
184
Snyder (1995), Christopher (1998: 135), Hutchings (1999), Caddy and Helou (2000: 1715), Planning &
Reporting (2001), Randall et al. (2002: 56), Sawyers (2003), Ballou (2004b: 335, 385-389), Kroll (2006).
185
Christopher (1998: 135).
186
Anupindi et al. (2006: 194f., 314, 335).
187
Electrical Wholesaling (1997), Faloon (1999), Franse (1999), Duvall (2000), Purchasing (2000), AFSAC
(2001), Gourmet Retailer (2002).
188
Verkoeijen and deHaas (1998).
189
190
Chopra and Meindl (2007: 321f.).
191
Benjaafar and Kim (2001: 13), Taylor (2004: 301ff.), Diruf (2005, 2007), Nonås and Jörnsten (2005:
521), Muckstadt (2005: 150ff.), Heil (2006), Sheffi (2006: 248, 2007), Yang and Schrage (2009: 837).
192
193
Diruf (2005, 2007), Nonås and Jörnsten (2005: 521), Heil (2006).
194
Cachon and Terwiesch (2009: 336ff.).
195
Schwarz (1989: 828), Pishchulov (2008: iii, 133).
196
Schwarz (1989: 828).
197
Sheffi (2007: 99f., 215-222).
198
Alfaro and Corbett (2003: 15), Neale et al. (2003), Reiner (2005: 434), Anupindi et al. (2006: 192, 194,
273), Sheffi (2006, 2007: 184-186), Brandimarte and Zotteri (2007: 30, 36, 38f.), Chopra and Meindl
(2007: 326ff.), Yang and Schrage (2009: 837), Bidgoli (2010: 21).
199
Benjaafar and Kim (2001: 13).
200
Zinn (1990: 12), Swaminathan (2001: 132), Simchi-Levi et al. (2008: 345ff.).
201
Swaminathan (2001: 132), Simchi-Levi et al. (2008: 345ff.).
25
(7) postponement, delayed (product) differentiation,202 process standardiza-

tion203, or “modularization strategies”204,
(8) capacity pooling and flexible manufacturing205,
(9) product pooling206, stock keeping unit (SKU) rationalization207, universal or
generic design208, or product standardization209, as well as
(10) product substitution210, “item substitution”211, interchangeability212, “de-
mand reshape”213, or “product commonalities”214.
Risk Pooling Methods
Building Blocks IP VP TS CO OS CC PM CP PP PS
DP 1 1 1 1 0 1 1 1 1 1
LP 0 0 1 0 1 0 0 0 0 0
Table 3.1: Risk Pooling Methods' Building Blocks
As highlighted in our definition in section 2.3, risk pooling (methods) may take advantage of
demand pooling (DP) and/or lead time pooling (LP). DP aggregates stochastic demands, so
that higher-than-average demands may balance lower-than-average ones.215 LP balances high-
er-than-average and lower-than-average lead times: A late-arriving order may be compensated
by an early-arriving one216, so that safety stock can be reduced, inventory availability in-
202
Benjaafar and Kim (2001: 13), Alfaro and Corbett (2003: 12), Neale et al. (2003), Fleischmann et al.
(2004: 93), Taylor (2004: 301ff.), Sheffi (2004: 95f., 100, 2006: 297, 2007), Hwarng et al. (2005: 2841,
2842), Reiner (2005: 434), Anupindi et al. (2006: 167, 169, 193), Enarsson (2006: 178), Heil (2006),
Brandimarte and Zotteri (2007: 39, 70), Chopra and Meindl (2007: 328f.), Pishchulov (2008: iii, 27, 133),
Sobel (2008: 172), Cachon and Terwiesch (2009: 341ff.), Yang and Schrage (2009: 837). Many authors
consider postponement and delayed (product) differentiation synonyms (e. g. Lee and Tang 1997, John-
son and Anderson 2000, Aviv and Federgruen 2001a: 512, 2001b: 578, Swaminathan and Lee 2003, Mat-
thews and Syed 2004, Anupindi et al. 2006: 193, Caux et al. 2006: 3244, Anand and Girotra 2007, Li et
al. 2007, Simchi-Levi et al. 2008: 190, 346, Graman and Sanders 2009, LeBlanc et al. 2009). At least
form postponement can be equated with delayed (product) differentiation and therefore the latter be con-
sidered a subform of postponement. For a distinction of these terms, which is not relevant for our further
considerations, please refer to Pishchulov (2008: 24-29).
203
204
Reiner (2005: 434).
205
Anupindi et al. (2006: 224), Graves (2008: 36), Cachon and Terwiesch (2009: 344ff.).
206
Cachon and Terwiesch (2009: 321, 330ff.).
207
Alfaro and Corbett (2003: 12), Neale et al. (2003), Sheffi (2006: 119, 2007), Sobel (2008: 172).
208
Pishchulov (2008: 17), Cachon and Terwiesch (2009: 330).
209
210
Anupindi et al. (2006: 192), Chopra and Meindl (2007: 324ff.), Yang and Schrage (2009: 837).
211
Benjaafar and Kim (2001: 13).
212
Sheffi (2006: 210, 2007).
213
Eynan and Fouque (2003, 2005).
214
Reiner (2005: 434).
215
Evers (1997: 55), Chen and Chen (2003).
216
Evers (1999: 122).
26
creased, or both217. Transshipments pool both lead times and demands, order splitting only the
former.218
Cachon and Terwiesch (2009: 336) consider consolidated distribution and delayed dif-
ferentiation types of LP: Lead times between the supplier and the retail stores in the direct-
delivery model are combined to a single lead time between the supplier and the distribution
center (DC) in the consolidated-distribution model.219 Actually (forecast) demands are
pooled over or during the outside-supplier lead time220. Thus it is likely that higher-than-
average and lower-than-average demands balance each other and inventory can be reduced
for a given service level (DP).
All risk pooling building blocks can reduce variability and thus inventory221 for a given
service level, increase the service level for a given inventory, or a combination of both222.
Table 3.1 can result in a classification according to the notation DP/LP that indicates
which risk pooling properties the risk pooling method of concern relies upon. A 1 indicates
that the respective property applies, a 0 the contrary. IP would be classified as a 1/0 – risk
pooling method.
3.1 Storage: Inventory Pooling
Inventory pooling is the combination of inventories and satisfying various demands from it in
order to reduce inventory holding and shortage costs through risk pooling.223 It is also called
vendor pooling224, product consolidation225, demand pooling226, pooling227, distributor integra-
tion228, location pooling229, or inventory centralization230. It can e. g. be achieved by invento-
ry231 or warehouse (system) centralization232 or selective stock keeping respectively specializa-
217
Evers (1997: 71, 1999: 122).
218
Evers (1997, 1999: 121).
219
220
Eppen and Schrage (1981), Schwarz (1989), Schoenmeyr (2005: 4), Gürbüz et al. (2007: 293).
221
Cf. Romano (2006: 320), Simchi-Levi et al. (2008: 48), Cachon and Terwiesch (2009: 325, 350), Bidgoli
(2010: 209).
222
Cf. Chopra and Meindl (2007: 336), Cachon and Terwiesch (2009: 325, 350).
223
Benjaafar and Kim (2001: 13), Kim and Benjaafar (2002: 12), Gerchak and He (2003: 1027), Benjaafar et
al. (2005), Anupindi et al. (2006: 191), Pishchulov (2008: 8, 17), Cachon and Terwiesch (2009: 322).
224
Monezka and Carter (1976: 207).
225
Benjaafar et al. (2004b: 73).
226
Benjaafar et al. (2004a: 1442, 2004b: 91).
227
Kemahloğlu-Ziya (2004), Bartholdi and Kemahloğlu-Ziya (2005), Benjaafar et al. (2005).
228
Simchi-Levi et al. (2008: 261, 263).
229
230
Monezka and Carter (1976: 207).
231
Benjaafar et al. (2004a: 1438).
232
Eppen (1979), Benjaafar and Kim (2001: 13), Taylor (2004: 304f.), Reiner (2005: 434), Heil (2006).
27
tion233. The latter strives to reduce inventory carrying cost treating products differently without
reducing the service level substantially. For example, products with a low turnover might be
stocked only at few locations due to cost considerations.234
Inventory pooling is e. g. considered in location and allocation decisions235, satisfying
both on-line and store demand236, speculative online exchanges237, airline revenue man-
agement238, for spare parts of airlines239 and of U.S. manufacturing companies240, and for
Saturn241, General Motors242, Intel243, and Airbus244.
Inventory pooling is a 1/0 – risk pooling method: Inventories are consolidated and sto-
chastic demands aggregated, as they are satisfied from the consolidated inventory245. Thus
demand variabilities may balance each other (demand pooling). The lead times to the sepa-
rate inventories are pooled to the lead time to the consolidated stock according to Cachon
and Terwiesch's (2009: 336, 339) notion of lead time pooling, but this actually is demand
pooling during the replenishment lead time. Inventory pooling does not pool lead times, so
that their variabilities may balance each other according to Evers (1997: 69ff., 1999: 121)
and Wanke and Saliby (2009: 678f.).
Inventory centralization can reduce expected costs in a cost minimization model246 or

increase profits and service level247 in a profit maximization model248 or in incentive-
compatible laboratory experiments in behavioral operations management (OM)249, even if
233
Anupindi et al. (2006: 191f.), Chopra and Meindl (2007: 322ff.)
234
Pfohl (2004a: 122f.).
235
Nozick and Turnquist (2001), Teo et al. (2001), Shen et al. (2003), Miranda and Garrido (2004), Hall
(2004), Croxton and Zinn (2005), Klijn and Slikker (2005), Shu et al. (2005), Snyder et al. (2007), Vi-
dyarthi et al. (2007), Naseraldin and Herer (2008), Ozsen et al. (2008), Taaffe et al. (2008), Lee and Jeong
(2009).
236
Bendoly (2004), Bendoly et al. (2007: 426).
237
Milner and Kouvelis (2007).
238
“Airline revenue management involves dynamically controlling the availability and prices of many dif-
ferent classes of tickets to maximize revenues” (Zhang and Cooper 2005: 415).
239
Hearn (2007), Burchell (2009), Cattrysse et al. (2009).
240
Carter and Monczka (1978).
241
Treece (1996).
242
Kisiel (2000).
243
Johnson (2005).
244
Aviation Week & Space Technology (2006).
245
Pishchulov (2008: 8, 17).
246
For example Eppen (1979), Chen and Lin (1989).
247
Eynan (1999).
248
For example Cherikh (2000), Lin et al. (2001a).
249
Ho et al. (2009).
28
locations differ in profitability (selling price and stockout costs)250, because “of risk pool-
ing over uncertainty in customer demand”251 and economies of scale252 and scope253.
Centralization (decreasing the number of warehouses) generally reduces inbound trans-
portation costs from the supplier to the warehouses because of a higher utilization of trans-
portation means (economies of density254), possibly the distances between the production
facility and the central warehouse(s)255, and the unit costs256 in the warehouse system due
to higher utilization (economies of scale). The systemwide total throughput is distributed
over a smaller number of warehouses and thus the average throughput per warehouse in-
creases.257 The warehouse costs per item (unit costs for space, product handling, and per-
sonnel) decrease through economies of scale.258 The fixed costs per order and per unit
holding costs usually diminish with centralization.259 Fixed warehouse costs, especially
personnel costs, are distributed over a larger warehouse throughput so that the unit costs
decrease with increasing utilization260 and the return on rationalization investments in more
productive low-cost methods increases (economies of scale)261. Variable warehouse costs,
particularly handling costs, diminish by mechanization or automation.262 However, increas-
ing mechanization or automation reduces flexibility263.
250
Eynan (1999: 38ff.) considers a central warehouse for a single product that supplies retailers from differ-
ent markets, which differ in product prices and stockout costs, according to the first come, first served
principle. Surprisingly the centralized system usually performs better than the decentralized one: Consi-
dering the whole system, possible cannibalization effects of low-price-market demands at the expense of
high-price-market ones are overcompensated.
251
Anupindi and Bassok (1999: 187), cf. Teo et al. (2001), Lim et al. (2002: 668), Snyder and Shen (2006),
Schmitt et al. (2008a, 2008b), Ho et al. (2009).
252
Lim et al. (2002: 668), Rabinovich and Evers (2003a). Economies of scale are decreasing unit costs with
increasing output (Adam 1979: 950, Gutenberg 1983, Busse von Colbe and Laßmann 1986: 197, Mono-
polkommision 1986: 231, Chandler 1990: 17, Bohr 1996: 375, Samuelson and Nordhaus 1998, Alisch et
al. 2004: 771).
253
Rabinovich and Evers (2003a). Economies of scope arise as cost synergy effects, if the combined produc-
tion of goods is cheaper than the separate one (Alisch et al. 2004: 771). In the literature among others the
portfolio effect is given as an economic reasoning for them: Diversification in different projects reduces
the total investment risk. Fritsch et al. (2001: 96) observe this effect with research and development in a
multi-product company, Bohr (1996: 382f.) with the general use of financial means.
254
Economies of density are cost advantages due to increasing utilization in a transportation system (Harris
1977, Caves et al. 1984: 472).
255
Fawcett et al. (1992: 37), Pfohl (1994: 141), Delfmann (1999: 193).
256
Unit costs are the costs that are defined by the distribution of total costs to the output to be produced
(Kistner and Steven 2002: 88).
257
Vahrenkamp (2000: 32).
258
Williams (1975), McKinnon (1989: 104).
259
Maister (1976: 132), McKinnon (1989: 102f.), Lim et al. (2002: 668).
260
Schulte (1999: 377), Vahrenkamp (2000: 34).
261
Paraschis (1989: 16), Pfohl (1994: 142), Vahrenkamp (2000: 34), Ihde (2001: 316).
262
Beuthe and Kreutzberger (2001: 249).
263
Stein (2000: 486), Ackerman and Brewer (2001: 231).
29
Centralization can lead to economies of massed reserves, if it is used to lower system-

wide safety stock through risk pooling and thus unit holding costs.264 The latter are further
reduced by less inventory loss and thus lower risk costs in the centralized system.265 Econ-
omies of massed reserves are cost savings due to centralized reserve holding with increas-
ing firm size.266 Among others Bowersox et al. (1986: 283ff.), Fawcett et al. (1992: 5),
Pfohl (1994: 141), and Delfmann (1999: 193) argue that in spite of constant basic invento-
ry and increasing in transit inventory total inventory is lowered by reducing safety stock in
the centralized system. Higher utilization of inbound transportation means with centraliza-
tion decreases the number of stock placements and removals.267 Therefore setups are re-
duced and larger loading and unloading equipment can be used saving time in the transpor-
tation system. Reducing idle time leads to a higher utilization in time and therefore to
economies of density using larger means of transportation.268
On the other hand, in a centralized logistics system outbound transportation costs may
be higher because of lower utilization269 and longer distances from the central ware-
house(s) to customers270 and thus perhaps the service level might be lower271. Centraliza-
tion may entail diseconomies of scale mostly in organization.272 With increasing size there
are no gains from rationalization anymore, but increasing transaction and coordination
costs and thus increasing warehouse unit costs in big warehouse systems.273
Das and Tyagi (1997) develop an optimization model for determining the optimal de-
gree of centralization as a tradeoff between inventory and transportation costs.
Inventory centralization games optimize and allocate the savings from a centralized
inventory, so that the participants' cooperation is maintained.274
264
Bowersox et al. (1986: 286), Paraschis (1989: 16), Vahrenkamp (2000: 34f.), Ihde (1976: 39f., 2001:
316).
265
Vahrenkamp (2000: 32).
266
Scherer and Ross (1990: 100).
267
Schulte (1999: 377).
268
Pfohl (1994: 142), Vahrenkamp (2000: 35).
269
In the medium term smaller transportation vehicles might be used to increase utilization (Fawcett et al.
1992: 5, Delfmann 1999: 193, Vahrenkamp 2000: 31).
270
Delfmann (1999: 193), Ballou (2004b: 573f.).
271
Schulte (1999: 377), Vahrenkamp (2000: 31), Lee and Jeong (2009: 1169), Wanke (2009: 122).
272
Chandler (1990: 25), Scherer and Ross (1990: 103f.), Bohr (1996: 385f.).
273
Pfohl (1994: 143), Delfmann (1999: 194).
274
Hartman et al. (2000), Anupindi et al. (2001), Slikker et al. (2001, 2005), Müller et al. (2002: 118f.),
Granot and Sošić (2003), Hartman and Dror (2005: 93), Özen and Sošić (2006), Ben-Zvi and Gerchak
(2007), Chen and Zhang (2007, 2009), Drechsel and Kimms (2007), Dror et al. (2008), Özen et al.
(2008), Chen (2009), Zhang (2009).
30
The risk pooling effect on (safety) stock levels evoked by inventory pooling or centrali-
zation can be quantified with the square root law (SRL), portfolio effect (PE), and inven-
tory turnover curve.
Maister's (1976) SRL states that the total system wide stock of n decentralized ware-
houses is equal to that of a single centralized one multiplied with the square root of the
number of warehouses n. There is some confusion about which part of inventory275 it can
be applied to and about its underlying assumptions276.
Nevertheless, it applies to regular stock277, if an economic order quantity (EOQ) order
policy is followed, the fixed cost per order and the per unit stock holding cost, demand at
every location, and total system demand is the same both before and after centralization.278
It holds for safety stock, if demands at the decentralized locations are uncorrelated279, the
variability (standard deviation) of demand at each decentralized location280, the safety fac-
tor (safety stock multiple)281 and average lead time are the same at all locations both before
and after consolidation, average total system demand remains the same after consolida-
tion282, no transshipments occur283, lead times and demands are independent and identical-
ly distributed random variables and independent of each other284, the variances of lead time
are zero285, and the safety-factor (kσ) approach is used to set safety stock for all facilities
both before and after consolidation286. It applies to total inventory, the sum of regular and
safety stock287, if the aforementioned assumptions of the SRL as applied to regular and
safety stock hold collectively. A formal proof is given in appendix B.
The savings in regular stock measured by the SRL stem from the assumption of constant
275
Heskett et al. (1974: 462), Ballou (2004b: 380). Professor emeritus Ronald H. Ballou, Weatherhead
School of Management, Case Western Reserve University, agreed with us in a personal correspondence
on May 15, 2007.
276
Sussams (1986: 9), Bowersox et al. (2002: 469), Coyle et al. (2003: 259f.), Ballou (2004b: 380), Chopra
and Meindl (2007: 320).
277
Regular or cycle stock is the inventory “necessary to meet the average demand during the time between
successive replenishments. The amount of cycle stock is highly dependent on production lot sizes, eco-
nomical shipment quantities, storage space limitations, replenishment lead times, price quantity discount
schedules, and inventory carrying costs” (Ballou 2004b: 331), and “affected by inventory policy” (Ballou
2004b: 379).
278
Maister (1976: 125, 127, 131).
279
Maister (1976: 132).
280
Maister (1976: 130).
281
Maister (1976: 132).
282
Evers and Beier (1993: 110), Evers (1995: 4).
283
Zinn et al. (1989: 2), Evers and Beier (1993: 110), Evers (1995: 4).
284
285
286
Maister (1976: 129).
287
Ballou (2004b: 380). Total inventory might also include pipeline, speculative, and obsolete stock (Ballou
2004b: 330f.), which is neglected here.
31
fixed costs per order, holding costs, and total demand for all locations before and after centrali-
zation. Thus, if an EOQ policy is followed, the total order fixed cost usually is lower because
of less orders and the inventory holding cost is lower due to a smaller total EOQ in the centra-
lized than in the decentralized system. For the proof please refer to appendix C. This is called
“EOQ cost effect” by Schwarz (1981: 147) and “order quantity effect” by Evers (1995: 5).
Savings in safety stock stem from the subadditivity of the square root (the square root of
the sum of the individual demand variances of the decentralized locations is usually less than
the sum of the standard deviations of demand of the decentralized locations), i. e. from ba-
lancing demand variabilities (demand risk pooling). Schwarz (1981: 147), Zinn et al. (1989:
3), and Evers (1995: 5) identify this as the portfolio effect.
Total cost savings from centralizing safety stock are probably larger than the ones from
cycle stock centralization288, because centralized cycle stocks have to be transported to the
customer eventually, so that extra transportation costs are probably high. Centralized safety
stocks only have to be transported to the customer in the less frequent case of a stockout, so
that additional expenses for premium transportation are relatively small. If both cycle and
safety stocks are centralized some warehouses might be closed and fixed costs saved.289
Although some researchers290 claim the SRL estimated real savings well, in 13291
of 14292 practical cases we reviewed it overestimated actual inventory savings often sig-
288
Maister (1976: 134), Evers (1995: 17).
289
Maister (1976: 134), Evers (1995: 17).
290
Pfohl (1994: 141). Sussams (1986: 10) gives an example where a company's savings in safety stock (cal-
culated with the standard deviation of demand of the decentralized and centralized depots) from reducing
the number of depots from five to one would only be 2.76 % higher than the ones predicted by the SRL,
although the variability (standard deviation) of demand at the decentralized depots is not the same, de-
mands are not uncorrelated as the correlation coefficients we calculated show, but all locations use the
same safety stock multiple (two times the standard deviation of demand). The SRL assuming uncorrelated
demands coincidentally predicts similar savings to the ones calculated on the basis of the actual standard
deviations in this case. Here the correlations of sales between the different depots balance each other
when demands are consolidated, so that Zinn et al.'s (1989: 3) portfolio effect (percent reduction of safety
stock due to centralization of correlated demands) of 57.25 % resembles the savings of the SRL of
55.27 %. However, as Zinn et al. (1989: 6, 10) show this need not be the case at all. Note that Sussams
(1986: 10) compares two theoretical savings calculated with the SRL and on the basis of calculated stan-
dard deviations and not the theoretical savings estimated by the SRL with the actual savings realized by
the company and his table contains mistakes. He concludes that the SRL “has been tested against 24 dif-
ferent situations, and in all cases there was close agreement between the theoretical and the practical re-
sults. It was concluded that, subject to the reservations concerning the normality of the demand pattern,
the square root law may be used with confidence to estimate differences in buffer stock requirements for
different configurations of depots”. However, it is not clear whether the savings theoretically predicted
with the SRL were compared to actually realized or calculated savings and whether the assumptions of
the SRL as applied to safety stock were fulfilled in the practical cases. Therefore Sussams' (1986: 10)
conclusion is not intersubjectively verifiable.
291
Newson (1978), Voorhees and Sharp (1978: 75f.). Ballou (1981: 148ff.) shows for several real companies
from different industries that the inventory savings predicted by the SRL are higher than the actual sav-
ings. This might be caused by joint ordering (to benefit from volume buying and shipping, which in-
creases inventory levels), forward buying, diversion from the EOQ pull policy for some products, use of
32
nificantly. This means its assumptions are not fulfilled293 in these cases or there were other
inventory-294 or service-related changes. It is difficult to isolate the effect of inventory cen-
tralization.295 The SRL shows the inventory necessary for a given service level in depen-
dence of the number of stocking locations, if its underlying assumptions are fulfilled. This
does not mean that reducing the number of warehouses automatically reduces inventory.
None of the aforementioned sources states whether the assumptions to the SRL were
fulfilled. Therefore actually no statements can be made about the accuracy of the SRL's
results. However, it seems that the SRL “tends to over-predict”296 actual inventory savings,
as its assumptions are not fulfilled: In a survey that we conducted in the winter of 2008
none of the eleven participating companies from different countries and industries fulfilled
all assumptions of the SRL when applied to safety stock (table D.1). They fulfilled at least
two (respondents 1, 4, and 5) and at the most seven (respondent 10) of these eleven as-
sumptions. Only one company (respondent 3) fulfilled all five assumptions necessary for
applying the SRL to regular stock. Two companies (respondents 4 and 11) did not fulfill
any of the assumptions of the SRL when applied to regular stock. Evers (1995: 17) already
remarked that the SRL is “based on numerous limiting assumptions which collectively are
not likely to be found in practice”, although Coyle et al. (2003: 259) call them “reasona-
ble”.
push distribution (where production considerations are more important than inventory replenishment
ones) or a stock-to-demand order policy (relates inventory levels directly to demand), multiple product
types with different service policies in the aggregated product group, no equal product service levels and
costs, and poor inventory management. According to Ashford (1997: 20), Nike Europe intended to reduce
its footwear storage locations from 20 to 2 and was predicted by the consultancy Touche Ross to save
11 % of inventory costs (calculated by us with the given data) instead of 68 % as predicted by the SRL.
Nike planned to reduce its apparel storage locations from 20 to 1 with predicted savings of 12 % (calcu-
lated by us) as opposed to 78 %. Close-outs were estimated to decrease by 33 % for footwear and 50 %
for apparel. It is not stated how the inventory savings were predicted and whether they were actually real-
ized. McKinnon (1997: 81), Watson and Morton (2000), Hammel et al. (2002). McKinnon (2003: 24)
considers two companies. Lovell et al. (2005), Progress Software Corporation (2007), clearpepper supply
chain consulting & education (2009). We interviewed the Supply Chain Manager of a U.S. wireless tele-
communication device production company with 6,000 employees. This manufacturer reduced the num-
ber of warehouses from 80 to 15 and thus reduced regular stock by 50, safety stock by 75, and total aver-
age inventory by 50 %. The SRL predicts inventory savings of 56.70 % and thus estimates the savings in
regular and total stock fairly well in this case. Table D.1 shows which assumptions of the SRL this res-
pondent 10 fulfilled.
292
A U.S. chemical company went from 100 to 89 warehouses. Inventory costs (including order processing
and warehousing costs) were reduced (projected by DISPLAN and later verified from accounting records)
by 76 % (compared to 6 % predicted by the SRL). These savings resulted not only from closing 11 ware-
houses, but also from reallocating demand among the remaining warehouses while maintaining the com-
pany's high level of customer service (Ballou 1979: 68).
293
Cf. Ballou (1981: 148ff.).
294
Ballou (1979: 68), McKinnon (1997: 81).
295
McKinnon (1997: 81).
296
McKinnon (1997: 81).
33
Researchers questioning Maister's (1976) SRL challenge its assumptions illogically297, do

make other assumptions and therefore arrive at different results298, or give no clear reason299.
Nonetheless, the other SRL- and PE-models compared in table D.2 might be applied
as their assumptions are fulfilled.
Eppen (1979) showed for the assumptions listed in table D.2 that the expected total sys-
tem inventory holding and penalty costs increase with the square root of the number of
warehouses.300 His research was extended to Poisson demand301, any (non-negative) de-
mand distributions with concave holding and penalty cost functions302, and to include costs
for the transportation of goods between locations with overage and underage in the centra-
lized system303.
Zinn et al.'s (1989) portfolio effect (PE) is a generalization of the SRL and shows the
reduction in aggregate safety stock by centralizing several warehouses' inventories in one
warehouse in percent for the assumptions listed in table D.2. The PE was used to estimate
the effect of postponement on safety stock304 and extended to include holding, transporta-
tion, facility investment, and procurement costs305, multiple consolidation points306, varia-
ble lead times307, both safety and cycle stocks (“consolidation effect”) as well as various
assumptions308, the effect of non-emergency309 and emergency transshipments310, and un-
equal demand variances and possibly unequal capacities of centralized locations311.
297
Durand (2007).
298
Das (1978), Evers and Beier (1993), Evers (1995).
299
Durand and Paché (2006).
300
Kemahlioğlu-Ziya (2004: 9) wrongly states that Eppen (1979) “is the first to model and analyze the
benefits of inventory centralization”.
301
Stulman (1987).
302
Chen and Lin (1989).
303
Chang and Lin (1991). They are extended by Chang et al. (1996), who consider delay supply product
cost.
304
Zinn (1990).
305
Mahmoud (1992: 203ff.).
306
Evers and Beier (1993).
307
Tallon (1993).
308
Evers (1995).
309
Evers (1996). In nonemergency transshipments a certain proportion of demand is always filled from other
locations' inventory (Evers 1996: 129). The PE model of Evers and Beier (1993) can be utilized to deter-
mine the percentage reduction in safety stocks from nonemergency transshipments without considering
the effect on cycle stocks and transportation costs (Evers 1996: 111).
310
Evers (1997). The PE model of Evers (1996) underestimates the benefits of an emergency transshipment
policy, as it does not adequately consider the number of locations, lead times, and desired fill rates (Evers
1997: 68ff.).
311
Tyagi and Das (1998: 197). Savings in aggregate safety stocks are maximal, if every centralized location
delivers the same portion of each decentralized location's demand. Centralization can be beneficial, even
if customers have unequal demand variances and more centralized locations are established than there are
customer ones (Tyagi and Das 1998: 201f.).
34
The inventory turnover curve shows average inventory in dependence of the inventory
throughput for a specific company. It can be constructed from a company's stock status
reports and be used to estimate the average inventory (not only safety stock as with the PE)
for any (planned) warehouse throughput (shipments from the warehouse or sales) without
the limitations of the SRL. Thus one can assess the effect of centralization, decentralization
(changing the number and/or size of warehouses), or combining warehouses on average
inventory as well as the performance of inventory management and control policies for a
specific company.312
3.2 Transportation
3.2.1 Virtual Pooling
Virtual pooling extends a company's warehouse or warehouses beyond its or their physical
inventory to the inventory of other own or other companies' locations313 by means of informa-
tion and communication technologies (ICT)314, drop-shipping315, and cross-filling316.
“[D]emand across [these locations] is pooled, which smoothes demand fluctuations”317. There-
fore inventory availability is improved for a given “inventory investment” or maintained with
less inventory.318 If virtual pooling entails cross-filling, it may pool lead times. However, this
corresponds to transshipments and is considered in the next section. Therefore virtual pooling
is a 1/0 – risk pooling method.
3.2.2 Transshipments
Lateral transshipments are monitored company internal or external product shipments on the
same level or echelon of the value or supply chain, e. g. among retailers.319 They are also
called intra-echelon transshipments320, stock transfer321, lateral (re)supply322, inventory shar-
ing323, stock redistribution324, virtual pooling325, or information pooling326.
312
Ballou (1981, 2000, 2004a, 2004b: 381f., 2005).
313
Caddy and Helou (2000: 1715), Planning & Reporting (2001), Kroll (2006).
314
Memon (1997), Christopher (1998: 135), Planning & Reporting (2001), SDM (2001), Frontline Solutions
Europe (2002), Electrical Wholesaling (2003), Mason et al. (2003), Fung et al. (2005), Kroll (2006), Ci-
oletti (2007), Cachon and Terwiesch (2009: 350, 469).
315
Planning & Reporting (2001), Randall et al. (2002, 2006), Netessine and Rudi (2006).
316
Ballou (2004b: 335, 385-389).
317
Randall et al. (2002: 56).
318
319
Lee (1987), Çömez et al. (2006), Herer et al. (2006: 185), Simchi-Levi et al. (2008: 239), Yang and
Schrage (2009: 838).
320
Burton and Banerjee (2005), Chiou (2008: 439).
321
Archibald et al. (1997).
35
Emergency or reactive327 lateral transshipments (ELT) transfer products from a loca-

tion with a surplus to a location that faces a stockout328. Yang and Schrage (2009: 838)
regard ELT as “location substitution”.
Preventive or proactive329 lateral transshipments are conducted, if a retailer anticipates
a stockout before demands are realized330. Lee et al. (2007) combine emergency and pre-
ventive lateral transshipments in their transshipment policy “service level adjustment”
(SLA).
In virtual lateral transshipments a capacitated manufacturing plant designates arriving
demands to be served by another, more remote one. This can help capacitated plants to
work together to better deal with random demands and save costs. In contrast to the source
plant of real lateral transshipments the one of the virtual lateral transshipment may have
negative inventory levels throughout the transshipment process.331
All of the above lateral transshipments must not be confused with cross-docking (re-
loading of goods at transfer points or hubs)332 or shipping goods from a warehouse to a
demand-satisfying location333, which is also called transshipments sometimes.
If locations from the same echelon cannot transship inventory to prevent a stockout, an
upper-level supplier can fill the demand via a cross-level, inter-echelon, or vertical
transshipment334. This might resemble virtual pooling carried out by drop-shipping335.
Emergency transshipments pool both demands across locations or retailers336 (by per-
mitting alternative locations to satisfy customer demands) and lead times (by providing
the whole system with the possibility of partial stock replenishments) and allow a company
322
Sherbrooke (1992), Yanagi and Sasaki (1992).
323
Anupindi et al. (2001), Grahovac and Chakravarty (2001), Zhao et al. (2005), Herer et al. (2006), Sošić
(2006), Chiou (2008: 427), Kutanoglu (2008).
324
Das (1975), Jönsson and Silver (1987a, 1987b), Bertrand and Bookbinder (1998), Tagaras and Vlachos
(2002), Hu et al. (2005).
325
Çömez et al. (2006).
326
Herer et al. (2006: 185), Özdemir et al. (2006a).
327
Herer et al. (2006: 185f.).
328
Lee (1987), Çömez et al. (2006), Herer et al. (2006: 185), Yang and Schrage (2009: 838).
329
Herer et al. (2006: 185f.).
330
Tagaras (1999), Lee et al. (2007). Preventive lateral transshipments are not useful in two-location periodic
review inventory systems with nonzero replenishment lead times and uncertain future material require-
ments and availability (Tagaras and Cohen 1992).
331
Yang and Qin (2007).
332
Campbell (1993), Hoppe and Tardos (2000), Petering and Murty (2009).
333
Köchel (2007).
334
Chiou (2008: 427, 439, 443).
335
Netessine and Rudi (2006).
336
Tagaras (1989), Tagaras and Cohen (1992), Evers (1997, 1999), Hong-Minh et al. (2000), Çömez et al.
(2006), Wanke and Saliby (2009), Yang and Schrage (2009: 837).
36
to remain close to customers337. They enable a company to exploit risk pooling338 or statis-
tical economies of scale339, i. e. “advantages that result from the pooling of uncertainty”340.
As a result (safety) stock341 and (inventory holding, shortage, backorder, or total sys-
tem) costs342 can be reduced for a given service level, which leads to leanness343, cus-
tomer service level (fill rates, product availability, or delivery time) can be improved
without raising inventory levels344, which enables agility or flexibility345, or a combina-
tion of both (increasing service while decreasing inventory levels)346, which is called “lea-
gility”347. Transshipments may decrease lost sales, the rejection rate of returns348, and rep-
lenishment lead times349 and improve revenues350 and performance351.
In a static stochastic demand system, transshipments balance stock levels at different lo-
cations through emergency stock transfers from one location with overage to another one
with underage. In a dynamic deterministic demand system, transshipments may save fixed
and variable replenishment costs352, if e. g. one location makes a larger replenishment or-
der to transship some of it to other locations353 (cf. section 3.3.1 on centralized ordering).
Inventory consolidations do not pool lead times, may close stock keeping installa-
tions near markets354, restructure the logistics network fundamentally, which is difficult to
change or undo at least in the short term355, and increase order cycle times and/or transpor-
337
Evers (1997, 1999), Wanke and Saliby (2009).
338
Tagaras and Vlachos (2002), Tagaras (1999), Dong and Rudi (2002, 2004), Çömez et al. (2006), Zhao
and Atkins (2009: 667).
339
Evers (1997, 1999).
340
Evers (1994: 51).
341
Jönsson and Silver (1987a: 224), Tagaras (1989), Diks and de Kok (1996: 378), Evers (1996), Mercer and
Tao (1996), Evers (1997: 71f.), Köchel (1998a: 190), Evers (1999), Kukreja et al. (2001), Herer et al.
(2002), Minner et al. (2003), Minner and Silver (2005).
342
Das (1975), Lee (1987), Robinson (1990), Evers (1996: 114), Köchel (1998a: 190), Alfredsson and Ver-
rijdt (1999), Evers (2001: 313), Grahovac and Chakravarty (2001), Kukreja et al. (2001), Herer et al.
(2002), Tagaras and Vlachos (2002), Daskin (2003), Minner et al. (2003), Kukreja and Schmidt (2005),
Minner and Silver (2005), Wong (2005), Çömez et al. (2006), Yang and Qin (2007), Chiou (2008: 428),
Kutanoglu and Mahajan (2009: 728).
343
Herer et al. (2002).
344
Krishnan and Rao (1965), Tagaras (1989), Chang and Lin (1991), Sherbrooke (1992), Evers (1996, 1997:
71f., 1999), Hong-Minh et al. (2000), Herer et al. (2002), Minner et al. (2003), Xu et al. (2003), Minner
and Silver (2005), Zhao et al. (2005), Chiou (2008: 427), Kutanoglu (2008: 356).
345
Herer et al. (2002), Zhao et al. (2008).
346
Evers (1997), Needham and Evers (1998), Evers (1999), Minner et al. (2003), Minner and Silver (2005),
Zhao et al. (2008: 400).
347
348
Ching et al. (2003).
349
350
Reyes and Meade (2006).
351
Wong (2005), Wong et al. (2007b: 1045), Chiou (2008: 429).
352
Herer and Rashit (1999: 525), Herer and Tzur (2003: 419).
353
Herer and Tzur (2003: 419).
354
Evers (1999: 122).
355
Domschke and Drexl (1996: 1), Klose and Stähly (2000: 434), Klose (2001: 3).
37
tation costs to customers356. The stockout probability for a given system inventory level is
lower in a decentralized system with emergency transshipments than in a centralized
one.357 Therefore these two systems are not generally equal as in Chang and Lin (1991)
with no replenishment times and no transshipment transportation costs.358
Transshipments are similar to order splitting359, but orders from facilities employing
transshipments are not necessarily placed at the same time, while split orders are360 and
order splitting only pools lead times361.
Transshipments may decrease customer service (increase the number of receipts per or-
der and/or order cycle times) and increase transportation or rebalancing362, transaction
(shipment documentation, receiving, handling and administration) costs363, the sending
location's cost of reordering the transshipped product from the supplier and its probability
of a stockout364, and daily average inventories365, and require willingness to share stock366.
Not all supply chain members may benefit from transshipments: they can increase
overall or retailer inventories and harm the distributor or individual retailers in a supply
chain, in which a manufacturer supplies integrated or autonomous retailers with a one-for-
one inventory policy via a central depot.367
Dong and Rudi (2002, 2004) find that if the wholesale price is endogenous, a single
manufacturer benefits from its retailers' transshipments, the more, the larger the risk pool-
ing effect. The identical (except in their normal demands) retailers in many situations are
worse off under transshipment due to a higher wholesale price, especially if the risk pool-
ing effect is large. Zhang (2005) extends Dong and Rudi's (2004) results under normal de-
mand to general demand distributions.
Shao et al. (2009) also conclude that a manufacturer and the decentralized retailers via
which he sells may be hurt by transshipments. If the manufacturer controls the transship-
ment price, he prefers selling through decentralized retailers. If the latter control the trans-
356
Evers (1999: 122).
357
Evers (1997: 69ff., 1999: 121).
358
Evers (1997: 69ff.).
359
Sculli and Wu (1981), Hayya et al. (1987), Sculli and Shum (1990).
360
Evers (1997: 75).
361
Evers (1999: 122, 132).
362
Jönsson and Silver (1987a: 224), Diks and de Kok (1996: 378), Evers (1997: 56), Evers (1999: 122),
Burton and Banerjee (2005: 169).
363
Evers (1999: 122, 2001: 312f.).
364
Evers (2001: 312f.).
365
Reyes and Meade (2006).
366
Evers (1999: 122).
367
Grahovac and Chakravarty (2001).
38
shipment price, they realize higher profits than a chain store and the manufacturer may
prefer selling via a chain store.
Most research deals with transshipments of goods or commodities between warehouses,
retailers, or production locations. Cohen et al. (1980) consider transfers of patients between
hospitals and capacity decisions. Lue (2006) introduces transshipments to chemical manu-
facturing systems with continuous material flows to maintain non-stop operations, since it
is costly to shut down and restart the production process.
Transshipments affect the ordering policy and should be considered in it.368 The majori-
ty of publications model transshipments and determine (optimal) replenishment order and
transshipment369, ordering370, stocking, or inventory control371, stocking or inventory con-
trol and transshipment372, production and transshipment373, and transshipment374 policies,
rules, heuristics, or algorithms under different assumptions.
Similarly to the inventory centralization games in section 3.1, Anupindi et al. (2001),
Granot and Sošić (2003), Sošić (2006), and Zhao et al. (2005) study the conditions (profit
allocations) that make retailers jointly share their residual supply/demand with the other
retailers in a manner that maximizes the additional profit in inventory sharing games.
Small incentives for transshipments can achieve the benefits of a full-inventory sharing
policy.375
Köchel (1998a) provides a survey on multi-location inventory models with lateral
transshipments, foremost on research conducted at the Technical University Chemnitz,
Germany concerning the coordination of ordering decisions of all locations. Chiou (2008)
and Paterson et al. (2009) present more general reviews of transshipment problems in
supply chain systems and inventory models with lateral transshipments.
In summary, transshipments are a 1/1 – risk pooling method.
368
Robinson (1990), Hu et al. (2005), Kukreja and Schmidt (2005).
369
For example Köchel (1975, 1977, 1982, 1988), Tagaras (1989), Arnold and Köchel (1996), Arnold et al.
(1997), Köchel (1998b), Herer and Rashit (1999), Bhaumik and Kataria (2006).
370
For example Robinson (1990), Tagaras and Vlachos (2002), Hu et al. (2005), Herer et al. (2006), Olsson
(2009).
371
For example Das (1975), Lee (1987), Tagaras and Cohen (1992), Kukreja et al. (2001), Jung et al. (2003),
Kukreja and Schmidt (2005), Wong (2005), Wong et al. (2005a, 2005b, 2007b), Gong and Yucesan
(2006), Lue (2006), Yang and Qin (2007), Kranenburg and Van Houtum (2009).
372
Diks and de Kok (1996), Archibald et al. (1997), Hu et al. (2008).
373
Zhao et al. (2008).
374
Tagaras (1999), Evers (2001), Axsäter (2003a), Minner et al. (2003), Burton and Banerjee (2005: 169),
Minner and Silver (2005), Wee and Dada (2005), Çömez et al. (2006), Zhao et al. (2006), Archibald
(2007), Lee et al. (2007), Archibald et al. (2009).
375
Zhao et al. (2005).
39
3.3 Procurement
3.3.1 Centralized Ordering
Centralized ordering376 places joint orders for several locations and later allocates the orders
(perhaps by a depot) to the requisitioners or distribution points in consolidated distribution
according to current demand information.377 This is also called pooling risk over (the) outside-
supplier lead time(s)378, “lead time pooling […] achieved by consolidated distribution”379,
“coordinated replenishment”380, “central ordering”381, or “centralized purchasing”382.
The allocation decision is postponed and stochastic demands can be treated in an aggre-
gate form until it is made. This reduces uncertainty and system stock “because of a portfolio
effect over the lead time from the supplier”383, “portfolio efficiencies”384, or “statistical
economies of scale”385. As explained at the beginning of this chapter, centralized ordering
does not entail lead time pooling, so that lower-than-average supplier lead times may not
balance higher-than-average ones. Consequently, centralized ordering is a 1/0 – risk pooling
method.
Compared to independent and individual or local386 ordering centralized ordering can lead
to economies of scale, i. e. it can take advantage of quantity discounts (“joint ordering ef-
fect”387) and save ordering388 and shipping costs389. It can make more frequent shipments
economical in comparison to direct delivery from the suppliers and thus reduce inventory
even further or increase the service level390.
If a warehouse, depot, or distribution center (DC) is introduced to allocate the stock391,
DC operating and extra transportation costs from the DC to the requisitioners or retailers are
incurred.392 A unit must travel a longer distance from the supplier to the retailer. However,
376
Eppen and Schrage (1981: 51), Erkip et al. (1990: 381), Ganeshan et al. (2007: 341), Gürbüz et al. (2007:
293).
377
Eppen and Schrage (1981), Cachon and Terwiesch (2009: 336-341).
378
Eppen and Schrage (1981), Schwarz (1989: 828), Schoenmeyr (2005: 4).
379
380
Gürbüz et al. (2007: 293).
381
Ganeshan et al. (2007: 341).
382
Anupindi et al. (2006: 149f.).
383
384
385
Eppen (1979: 498), Eppen and Schrage (1981: 52).
386
387
388
Hu et al. (1997, 2005: 34), Gürbüz et al. (2007: 293).
389
Gürbüz et al. (2007: 293), Cachon and Terwiesch (2009: 341).
390
391
Eppen and Schrage (1981), Schwarz (1989: 828), Cachon and Terwiesch (2009: 336ff.).
392
McGavin et al. (1993: 1094), Cachon and Terwiesch (2009: 341).
40
the holding cost for each unit of inventory at the DC is probably lower than at the retail
stores.393
The value of risk pooling through holding inventory at the warehouse, depot, or DC (“de-
pot effect”394) and using this inventory “between system replenishments to ‘rebalance’ retail-
er inventories which have become ‘unbalanced’ due to variations in individual retailer de-
mands” (between replenishment risk pooling)395 can be substantial396 (cf. inventory pooling).
Schwarz et al. (1985) and Badinelli and Schwarz (1988) find it is not significant, perhaps
because the model of Deuermeyer and Schwarz (1981) that they use does not allow the
warehouse to balance retailer inventory levels, if system stocks are low397.
3.3.2 Order Splitting
Order splitting is simultaneously partitioning a replenishment order into multiple orders with
multiple suppliers398 or into multiple deliveries (scheduled-release)399. The single order and
thus its lead time are split into multiple orders or deliveries and their lead times, so that the
variabilities of these lead times may balance each other400. Thus safety stock needed for a giv-
en service level (inventory availability, expected number of backorders or shortage cost) can be
reduced, service level increased for a given safety stock level, or both.401 It can also reduce
cycle stock due to successive deliveries of smaller split orders.402 Consequently, order splitting
only pools lead times, not demands403 and constitutes a 0/1 – risk pooling method.
Disadvantages of order splitting might include no “quantity and transportation dis-
counts” because of smaller orders, a higher “administrative effort”, which electronic data
interchange might mitigate, and no “long-term partnership with a single supplier”, but mul-
tiple “suppliers may ensure competitive pricing and other favorable terms”.404
Thomas and Tyworth (2006: 246f., 2007: 170f.) critically review the literature on order
splitting: On the one hand, the literature assesses the effect of order splitting on the distri-
393
394
395
McGavin et al. (1993: 1093), cf. Schwarz (1989: 830).
396
Jackson and Muckstadt (1984a, 1984b, 1989), Jönsson and Silver (1987a, 1987b), Jackson (1988),
McGavin et al. (1993: 1104).
397
McGavin et al. (1993: 1093).
398
Evers (1999: 123), Thomas and Tyworth (2006: 245), Qi (2007).
399
Hill (1996), Chiang (2001), Mishra and Tadikamalla (2006), Thomas and Tyworth (2007: 188).
400
Evers (1999: 122).
401
Evers (1997: 71), Evers (1999: 123), Thomas and Tyworth (2006: 245f.).
402
403
Evers (1999: 123f.).
404
Evers (1999: 123).
41
bution of effective lead times405 and safety stock carrying and shortage costs with statis-
tics.406 On the other hand, it compares the average total cost (inventory holding, ordering,
shortage, and purchase cost) of order splitting with the one of not order splitting407 and
finds that it reduces cycle stock408. Order splitting and the just-in-time inventory strategy
are analyzed by Pan and Liao (1989), Ramasesh (1990), Hong and Hayya (1992), Hong et
al. (1992), Kelle and Miller (2001), and Ryu and Lee (2003). Simultaneously splitting an
order among suppliers does not decrease the sum of in-transit and cycle stock, but only
total safety stock in the system409 and increases ordering410 and shipping costs411. Most
researchers neglect transportation economies of scale, the size of transportation costs com-
pared to inventory costs, in-transit inventory412, the probability of lead time correlation413,
order dependent supply lead times and unit purchasing prices, and the derived disadvan-
tages and advantages of order splitting414.
Options other than simultaneously splitting replenishment orders among several suppli-
ers might be more promising:415 Delivering an order in sequential shipments can reduce
demand and production variability because of advance demand and delivery informa-
tion.416 Orders can also be split between a fast, reliable and a cheaper, less reliable mode or
supplier in order to take advantage of cost-performance differences in transportation mod-
es417 or between capacitated suppliers418. Lead time variability and freight rates could be
405
“The effective lead time in this case will be that of the minimum of the set of random variables
representing the lead time of each supplier” (Sculli and Wu 1981: 1003).
406
Sculli and Wu (1981), Hayya et al. (1987), Kelle and Silver (1990a, 1990b), Sculli and Shum (1990), Pan
et al. (1991), Ramasesh (1991), Fong (1992), Fong and Ord (1993), Guo and Ganeshan (1995), Fong and
Gempeshaw (1996), Fong et al. (2000), Geetha and Achary (2000), Kelle and Miller (2001), Mishra and
Tadikamalla (2006).
407
Ramasesh (1990), Ramasesh et al. (1991), Hong and Hayya (1992), Lau and Zhao (1993), Ramasesh et
al. (1993), Chiang and Benton (1994), Lau and Lau (1994), Lau and Zhao (1994), Gupta and Kini (1995),
Chiang and Chiang (1996), Mohebbi and Posner (1998), Ganeshan et al. (1999), Sedarage et al. (1999),
Tyworth and Ruiz-Torres (2000), Chiang (2001), Ghodsypour and O'Brien (2001), Ryu and Lee (2003).
408
Moinzadeh and Nahmias (1988), Pan and Liao (1989), Ramasesh (1990), Sculli and Shum (1990), Pan et
al. (1991), Ramasesh et al. (1991), Hong and Hayya (1992), Hong et al. (1992), Zhou and Lau (1992),
Lau and Zhao (1993), Ramasesh et al. (1993), Chiang and Benton (1994), Lau and Zhao (1994), Gupta
and Kini (1995), Chiang and Chiang (1996), Hill (1996), Ganeshan et al. (1999), Chiang (2001), Ghodsy-
pour and O'Brien (2001), Ryu and Lee (2003).
409
410
Sajadieh and Eshghi (2009).
411
412
Thomas and Tyworth (2006: 254, 2007: 188).
413
Minner (2003: 273), Thomas and Tyworth (2006: 254, 2007: 188).
414
Sajadieh and Eshghi (2009: 3272).
415
416
Hill (1996), Janssen et al. (2000), Minner (2003: 276), Thomas and Tyworth (2006: 255).
417
Ganeshan et al. (1999), Minner (2003).
418
Qi (2007).
42
reduced by stable, periodic, long-term transportation commitments.419 Using emergency

transshipments instead of order splitting to reduce variation is attractive in many situations,
not only because order splitting is contrary to current theory and practice of vendor rela-
tions and lean production.420
3.4 Production
3.4.1 Component Commonality
Component commonality or part standardization421 designs products that share parts or com-
ponents422. These common components can be used for several products.423
Component commonality is a 1/0 – risk pooling method: Demand for the individual
components is aggregated (pooled424) to the demand for the (fewer) generic common com-
ponent(s).425
This may426 reduce the variability of demand (quantity and timing uncertainty) and the
required (safety) stock for any constant service level427, increase the service level for a
given inventory, or lead to a combination of both, inventory reduction and service in-
crease428 due to risk pooling429. Furthermore, component commonality can lower setup,
ordering, inventory holding430, manufacturing431, logistics432, and part costs433 due to econ-
419
Thomas and Hackman (2003), Henig et al. (1997).
420
Evers (1999: 133).
421
Wacker and Treleven (1986: 219), Swaminathan (2001: 129), Simchi-Levi et al. (2008: 345).
422
Srinivasan et al. (1992), Jönsson et al. (1993), Meyer and Lehnerd (1997), Ma et al. (2002), Mirchandani
and Mishra (2002), Kim and Chhajed (2001), Labro (2004), Van Mieghem (2004), Ashayeri and Selen
(2005), Chew et al. (2006), Humair and Willems (2006). Swaminathan (2001: 131) and Simchi-Levi et al.
(2008: 348) also consider commonality in procurement standardization, which exploits commonality in
part and equipment purchasing. Demand can be pooled across a large variety of end products, if they are
produced on similar machines and/or use common components. Procurement standardization is most ef-
fective, if the product is nonmodular and the process modular.
423
Grotzinger et al. (1993: 524).
424
425
Dogramaci (1979: 129), Guerrero (1985: 409), Vakharia et al. (1996: 15), Kreng and Lee (2004), Labro
(2004: 363), Kulkarni et al. (2005: 247), Graman and Magazine (2006: 1074), Bidgoli (2010: 21).
426
Component commonality does not always reduce inventory in a dynamic inventory system with lead
times under some common allocation rules (Song and Zhao 2009: 493).
427
Roque (1977), Collier (1981: 95, 1982: 1303), McClain et al. (1984), Baker et al. (1986), Gerchak and
Henig (1986, 1989), Wacker and Treleven (1986: 219), Gerchak et al. (1988).
428
Bagchi and Gutierrez (1992: 824f., 829), Srinivasan et al. (1992), Grotzinger et al. (1993: 525f.),
McDermott and Stock (1994: 65), Eynan (1996), Eynan and Rosenblatt (1996), Thonemann and Bran-
deau (2000), Hillier (2002a, 2002b), Thomas et al. (2003), Labro (2004: 363), Mohebbi and Choobineh
(2005: 481), Chopra and Meindl (2007: 326ff.).
429
Srinivasan et al. (1992: 26f.), Grotzinger et al. (1993: 524), Swaminathan (2001: 129), Labro (2004: 363),
Chew et al. (2006), Simchi-Levi et al. (2008: 345).
430
Dogramaci (1979: 129), Collier (1981: 95, 1982: 1303), Guerrero (1985: 402), Thonemann and Brandeau
(2000), Swaminathan (2001: 129, 131), Hillier (2002b: 570), Ma et al. (2002: 524), Mirchandani and
Mishra (2002), Mohebbi and Choobineh (2005: 473), Simchi-Levi et al. (2008: 345, 348).
43
omies of scale434 or order pooling435, as well as product design costs436 and design and
manufacturing time437. The order-pooling benefit may be more important than the risk-
pooling one.438 Component commonality increases the survival probability of start-up
companies439 and possibly revenue440.
Total stock of product-specific components may increase though441 depending on the
used service-level measure442. Although component commonality may reduce the average
work load, it may increase workload variability and work-in-process inventory variabili-
ty.443
Excessive part commonality can reduce product differentiation, so that less expensive
customization options might cannibalize sales of more expensive parts.444 While customer
valuation for the high-value product and the chargeable price may be reduced, the ones for
the low-value one may be increased by using commonality in vertical product line exten-
sions.445
Sometimes, it is necessary to redesign products and/or processes to achieve commonali-
ty.446 This may result in a smaller and more economical set of components though.447 The
431
Collier (1981: 95), Hayes and Wheelwright (1984: 229-274), Walleigh (1989), Kim and Chhajed (2001:
219).
432
Kim and Chhajed (2001: 219).
433
Collier (1982: 1303).
434
Collier (1982: 1303), Kim and Chhajed (2001: 219), Swaminathan (2001: 129, 131), Simchi-Levi et al.
(2008: 345, 348).
435
Hillier (2002b: 570).
436
Desai et al. (2001).
437
Maskell (1991: 184f.), Berry et al. (1992), McDermott and Stock (1994: 65), Sheu and Wacker (1997),
Mirchandani and Mishra (2002).
438
Hillier (2002b: 570).
439
Thomas et al. (2003).
440
Component commonality allows to produce more of a higher-margin product instead of a low-margin
one, if demand exceeds capacity. Therefore it may be optimal even for perfectly positively correlated de-
mands (no risk-pooling benefit), if products have sufficiently different margins (“revenue-maximization
option”) (Van Mieghem 2004: 422).
441
Baker et al. (1986), Gerchak and Henig (1986: 157), Gerchak et al. (1988), Gerchak and Henig (1989:
61), Van Mieghem (2004: 423).
442
Gerchak et al. (1988).
443
Guerrero (1985: 409), Vakharia et al. (1996: 3), Ma et al. (2002).
444
Ulrich and Ellison (1999), Kim and Chhajed (2000, 2001: 219), Desai et al. (2001), Krishnan and Gupta
(2001), Ramdas and Sawhney (2001), Simpson et al. (2001), Swaminathan (2001: 131), Heese and Swa-
minathan (2006), Simchi-Levi et al. (2008: 348).
445
Kim and Chhajed (2001: 219).
446
Bagchi and Gutierrez (1992: 817), Swaminathan (2001: 129), Ma et al. (2002: 536), Simchi-Levi et al.
(2008: 345).
447
Whybark (1989).
44
common component may be more expensive than the unique components it replaces.448
Still component commonality may reduce total cost.449
Using commonality as backup safety stock, if unique parts are not available, is better
than no commonality or pure commonality, where only common parts are used.450 Pure
commonality is not optimal unless it is free or there are high fixed product and process
redesign or complexity costs for procuring and handling two inputs.451
A component-mismatch problem due to demand uncertainty for end products may
arise.452 The equal-fractile allocation policy453 that allocates the components so that all
products have an equal probability of being out of stock at the end of the period can help to
reduce the required safety stock for a given target service level. However, it may keep
amply available components at the component level for the product with the highest de-
mand variability (“worst-case product”), although they may be employed to manufacture
other end products. This can be avoided by modifying the policy accordingly.454
Although some publications assert that cost decreases with commonality in general,
there are contradictory claims on the effect of commonality on various cost elements and
commonality's impact on total cost cannot be determined in general yet.455
Wazed et al. (2008) review the literature on component commonality. For a review of
advantages and disadvantages of component commonality in manufacturing and degree of
commonality indices in designing new or assessing existing product families please refer to
Wazed et al. (2009). Boysen and Scholl (2009) develop a general solution framework for
component-commonality problems.
3.4.2 Postponement
Postponement in general means delaying a decision in production, logistics, or distribution in

order to be able to use more accurate information because of a shorter forecast period and an
aggregate forecast456, especially in industries with high demand uncertainty,457 and commit-
448
Hillier (2002a, 2002b), Van Mieghem (2004: 419), Zhou and Grubbström (2004), Mohebbi and Choobi-
neh (2005: 473).
449
Van Mieghem (2004: 419), Hillier (2002a, 2002b).
450
Hillier (2002a).
451
452
Baker (1985), Chew et al. (2006: 240).
453
Cf. Eppen and Schrage (1981), Bollapragada et al. (1998).
454
Chew et al. (2006: 239f., 248).
455
Labro (2004: 358, 366).
456
Aggregate forecasts usually are more accurate than disaggregate ones (Lawrence and Zanakis 1984: 25,
Neumann 1996: 9f., Swaminathan 2001: 126f., Sheffi 2004: 93f., Alicke 2005: 44, Nahmias 2005: 55,
Anupindi et al. 2006: 168, 187, Donnellan et al. 2006: x, Chopra and Meindl 2007: 188f., Simchi-Levi et
al. 2008: 190, 194, 345, Shah 2009: 166, Bretzke 2010: 77f., Schnuckel 2010: 152) because of statistical
45
ting resources rather to demand than to a forecast458. The opposite strategy of holding finished
goods at locations close to customers in anticipation of sales is called speculation.459
Postponement strategies can be applied to form, time, and place.460 This can include
product development461, purchasing462, order(ing)463, fulfillment assignment464, produc-
tion465, manufacturing466, assembly467, packaging468, labeling469, delivery470, and pricing
postponement471. Form postponement delays product finalization or differentiation until
after customer orders have been received.472 Time postponement delays the forward
movement of stock.473 Rabinovich and Evers (2003b: 36, 41f.) consider emergency trans-
shipments and inventory centralization forms of time postponement. Place postponement
keeps inventories in centralized locations until customer orders are received.474 Logistics or
geographic postponement combines time and place postponement.475 It delays the transpor-
tation to individual market areas.476 Pull postponement moves the decoupling point477 up-
balancing effects, statistical economies of scale, or risk pooling (Bretzke 2010: 77f.). For further explana-
tions please refer to e. g. Neumann (1996: 9f.), Sheffi (2004: 93), Alicke (2005: 44), Nahmias (2005: 55),
Donnellan et al. (2006: x), Simchi-Levi et al. (2008: 60), and Schnuckel (2010: 152).
457
Pfohl (1994: 143), Swaminathan and Tayur (1998), Swaminathan (2001: 129f.), Sheffi (2004: 95f., 100),
Anupindi et al. (2006: 192f.), Chopra and Meindl (2007: 362), García-Dastugue and Lambert (2007:
57f.), Piontek (2007: 86f.), Simchi-Levi et al. (2008: 346), LeBlanc et al. (2009: 19).
458
Bucklin (1965), Van Hoek (2001: 161).
459
Bucklin (1965).
460
Van Hoek et al. (1998: 33).
461
Yang et al. (2004: 1052), Piontek (2007: 86).
462
Yang et al. (2004: 1052).
463
Granot and Yin (2008), Chen and Lee (2009), Li et al. (2009).
464
Allowing online sales to accumulate and postponing assigning them to fulfillment sites can reduce hold-
ing, backorder, and transportation costs (Mahar et al. 2009: 561).
465
Yang et al. (2004: 1052), Anupindi and Jiang (2008: 1876).
466
Zinn and Bowersox (1988), Huang and Lo (2003).
467
Zinn and Bowersox (1988).
468
Zinn and Bowersox (1988), Howard (1994), Graman and Magazine (1998), Cholette (2009), Graman
(2010).
469
Zinn and Bowersox (1988), Cholette (2009).
470
Anupindi and Jiang (2008: 1876).
471
Van Mieghem and Dada (1999), Boone et al. (2007: 597), Granot and Yin (2008).
472
Bowersox and Closs (1996), Lee (1998), Van Hoek (2001: 167), Rabinovich and Evers (2003b: 33), Li et
al. (2007: 31), Harrison and Skipworth (2008), Wong et al. (2009).
473
Zinn and Bowersox (1988), Bowersox and Closs (1996), Van Hoek et al. (1998: 33), Rabinovich and
Evers (2003b: 33), Nair (2005: 449).
474
Bowersox and Closs (1996), Van Hoek et al. (1998: 33), Rabinovich and Evers (2003b: 33), Nair (2005:
449).
475
Pfohl (1994: 145), Bowersox and Closs (1996: 473), Lee (1998), Van Hoek (1998a), Van Hoek et al.
(1998: 33), Özer and Xiong (2008).
476
Pfohl (1994: 145).
477
Up to the decoupling point, order-penetration-point, variant determination point, freeze point (Piontek
2007: 86f.), or point of (product) differentiation (PoD) (Lee 1996, Lee and Tang 1997, Meyr 2003) stan-
dardized products are made to stock customer-anonymously based on sales forecasts (push strategy). Af-
ter an order has been received, the standardized products are transformed into various variants customer-
individually (make-to-order pull strategy) (Piontek 2007: 86f., Harrison and Skipworth 2008).
46
stream to assemble to order478. Waller et al. (2000: 138) consider upstream postponement,
e. g. delaying ordering of raw materials, and downstream postponement, e. g. distribution
or place postponement.
Postponement allows to ship a single common generic product longer down the supply
chain and change it to individual products (differentiate it) more responsively according to
more recent demand information later.479 On the preceding levels of the supply chain the
demands for the individual products are aggregated to the demand for the generic prod-
uct (demand pooling480), which fluctuates less, since the stochastic fluctuations of the indi-
vidual demands balance each other to a certain extent because of the risk pooling effect.481
Postponement constitutes a 1/0 – risk pooling method.
Furthermore, inventories of the common intermediate product can function as a com-
mon buffer.482 A learning effect may arise, if forecasts can be improved on the basis of
observed sales data.483 Thus the total inventory holding costs and stockouts can be reduced
and customer service level484 and profits increased485 by better matching supply and de-
mand486.
Besides this risk and cost reduction, postponement can lead to economies of scale, syn-
ergistic effects, delayed increase in costs in the value chain and thus decreased capital
commitment, and higher flexibility or leagility.487 However, it is usually necessary to rede-
sign488 the product specifically for delayed differentiation and to modify the order of man-
ufacturing steps (resequence or operations reversal)489. This may decrease production vo-
lume variability though.490 Time-based postponement at the company level can increase
478
Lee (1998).
479
Lee (1996), Aviv and Federgruen (2001b: 579).
480
481
Lee and Tang (1997: 52), Aviv and Federgruen (2001a: 514, 2001b: 579), Alfaro and Corbett (2003: 12,
15), Piontek (2007: 87), Dominguez and Lashkari (2004: 2113), Anupindi et al. (2006: 192f.), Caux et al.
(2006), Jiang et al. (2006), Cholette (2009).
482
Weng (1999), Aviv and Federgruen (2001b: 579).
483
Aviv and Federgruen (2001b: 579).
484
Lee (1996), Lee and Tang (1997), Jiang et al. (2006), Davila and Wouters (2007), Li et al. (2008, 2009),
LeBlanc et al. (2009: 19), Mahar and Wright (2009), Wong et al. (2009).
485
Jiang et al. (2006), Chopra and Meindl (2007: 364), Cholette (2009).
486
487
Aviv and Federgruen (2001b: 579), Herer et al. (2002), Piontek (2007: 86f.).
488
Swaminathan (2001: 129f.), Kim and Benjaafar (2002: 12), Yang et al. (2005b: 994), Davila and Wouters
(2007: 2245), Shao and Ji (2008), Simchi-Levi et al. (2008: 346).
489
Lee and Tang (1997: 40, 1998), Kapuscinski and Tayur (1999), Jain and Paul (2001), Swaminathan
(2001: 129f.), Shao and Ji (2008), Simchi-Levi et al. (2008: 346).
490
Jain and Paul (2001).
47
supply chain inventory, as other members of the supply chain may be forced to use more
speculation, i. e. hold more inventories.491
Postponement reduces risk associated with order mix and order quantity, but may in-
crease stockouts or lost sales due to longer lead, delivery, or order cycle times.492 Specula-
tion provides a high availability and flexibility because of short lead times and few stock-
outs.493 Decreasing (increasing) speculation and increasing (decreasing) postponement
increases (decreases) transportation costs and decreases (increases) inventory holding
costs.494
Postponement and speculation can be combined installing an intermediary stage be-
tween the suppliers and the buyers close to the buyers and holding uncommitted, anticipa-
tory inventory of the suppliers ready to be shipped on request495 or “postponing only the
uncertain part of the demand and producing the predictable part at a lower cost without
postponement” in “tailored postponement”496. “The combined principle of postponement-
speculation”497 may lower both inventory risk by centralization and order fulfillment times
and lost sales.498 Internet retailers, for instance, depend more on both inventory location
speculation (in-stock inventory) and postponement (drop-shipping) to fulfill their orders as
their market share and product popularity increase.499
As products are only delivered after a customer order is received, in general postpone-
ment leads to a centralized logistics system.500 Speculation rather has a decentralizing
effect on the logistics system501 and finished products inventory is stored as close as possi-
ble to customers. At the same time speculative processes lead to consolidated flows of
goods because of higher order quantities.502 Consolidated flows of goods support the ten-
dency towards centralized logistics systems.503 These opposite movements are no contra-
491
García-Dastugue and Lambert (2007).
492
Zinn and Bowersox (1988: 126), Van Hoek (2001: 161), Rabinovich and Evers (2003b: 35), Yang et al.
(2005b: 994), Graman and Magazine (2006: 1078), Pishchulov (2008: 13).
493
Bucklin (1965: 27), Delfmann (1999: 195).
494
Bucklin (1965: 27), Zinn and Bowersox (1988: 126), Graman and Magazine (1998), Delfmann (1999:
195), Van Hoek (2001: 161).
495
Bucklin (1965: 31).
496
497
Bucklin (1965: 28).
498
Bucklin (1965: 31), Pishchulov (2008: 27).
499
Bailey and Rabinovich (2005).
500
Shapiro and Heskett (1985: 53), Zinn and Bowersox (1988: 126), Pfohl (1994: 145), Pagh and Cooper
(1998: 14), Van Hoek (1998a: 95), Van Hoek et al. (1999b: 506), Nair (2005: 458).
501
Shapiro and Heskett (1985: 53), Klaas (2002: 154).
502
Pagh and Cooper (1998: 14), Delfmann (1999: 195), Klaas (2002: 157).
503
Pfohl et al. (1992: 95), Kloster (2002: 122).
48
diction, but rather proof of the complex interdependencies between process and structure
variables in logistics systems.504
In practice there is a trend towards customer order oriented logistics systems505 or post-
ponement as a logical consequence to centralization506, increasing uncertainty in buyer
markets, increasing product variety, individualization of demands, and regional expansion
of supply in a globalized economy507.
Postponement is analyzed by Eppen and Schrage (1981) in the steel industry, Heskett
and Signorelli (1984) at Benetton, Fisher and Raman (1996) at Sport Obermeyer, Feitzin-
ger and Lee (1997) at HewlettPackard, Magretta (1998), Van Hoek (1998b), and Kumar
and Craig (2007) at Dell, Inc., Van Hoek (1998b) with the SMART car, Battezzati and
Magnani (2000) for industrial and fast moving consumer goods in Italy, Brown et al.
(2000) at a semiconductor firm, Chiou et al. (2002) in the Taiwanese IT industry, Huang
and Lo (2003) in the Taiwanese desktop personal computer industry, Dominguez and
Lashkari (2004) at a major household appliance manufacturer in Mexico, Caux et al.
(2006) in the aluminum-conversion industry, Davila and Wouters (2007) at a disk drive
manufacturer, Cholette (2009) in wine distribution, ElMaraghy and Mahmoudi (2009) in
the optimal location of nodes of a global automobile wiper supply chain considering cur-
rency exchange rates and the optimal modular product structure, Kumar et al. (2009) at 3M
Company, Kumar and Wilson (2009) in off-shored manufacturing, and Wong et al. (2009)
in terms of the optimal differentiation point positioning and stocking levels.
Van Hoek (2001) reviews the literature on postponement dating back to 1965 and puts
it in a systematic framework. Boone et al. (2007) review postponement literature published
from 1999 to 2006 and conclude research should be extended to non-manufacturing post-
ponement, investigate the slow rate of postponement adoption among practitioners, and
continue assessing the relationship between postponement and uncertainty.
3.4.3 Capacity Pooling
Capacity pooling is the consolidation of production508, service509, transportation510, or invento-

ry capacities of several facilities.511 Without pooling every facility fulfills demand just with its
504
Klaas (2002: 157f.).
505
Van Hoek et al. (1999b: 506).
506
Van Hoek (1998a: 95).
507
Ihde (2001: 36).
508
Plambeck and Taylor (2003), Iyer and Jain (2004), Jain (2007), Simchi-Levi et al. (2008: 281).
509
Cachon and Terwiesch (2009: 149ff., 325, 349, 467).
510
Masters (1980: 71), Evers (1994), Ihde (2001: 33), Chen and Chen (2003), Chen and Ren (2007).
49
own capacity. With pooling demand is aggregated and fulfilled by a single (perhaps virtually)
joint facility.512 If demand is stochastic, a higher service level can be attained with the same
capacity or the same service level can be offered with less capacity.513 It is also advantageous,
if there are economies of scale in obtaining capacity or satisfying demand.514 Capacity pooling
may pool supplier lead times, if the capacities receive separate deliveries from the suppliers
and provide the whole system with the possibility of partial stock replenishments515. However,
this is considered under stock sharing (transshipments). Consequently, capacity pooling is re-
garded as a 1/0 – risk pooling method.
It is predominantly associated with combining manufacturing capacity516 and thus creat-
ing manufacturing flexibility517. We adopt this view in the following and subsume pooling
of inventory capacity under inventory pooling. Manufacturing flexibility means that a plant
is capable of producing more than one product. With no flexibility each plant can only
produce one product, with total flexibility every plant can produce every product (as in
manufacturing postponement). Flexibility allows production shifts to high selling products
to avoid lost sales.518
Capacity pooling can increase effective capacity to serve more demand, which leads to
higher expected sales, profits, and capacity utilization519 or lower manufacturing flow
time.520 Aggregating demands which were served by individual capacities before and are
now served with pooled capacity can reduce demand variability and thus the demand-
capacity mismatch cost. However, installing flexibility is expensive521. If demands are of
different variability, pooling production capacities may increase inventory costs for the
low-demand-variability facility522 or total cost523.
511
Anupindi et al. (2006: 223), Yu et al. (2008: 1), Cachon and Terwiesch (2009: 463).
512
Yu et al. (2008: 1).
513
Anupindi et al. (2006: 224), Yu et al. (2008: 1).
514
Yu et al. (2008: 1).
515
Cf. Evers (1997, 1999), Wanke and Saliby (2009).
516
Plambeck and Taylor (2003), Iyer and Jain (2004), Jain (2007), Simchi-Levi et al. (2008: 281).
517
Upton (1994, 1995), Jordan and Graves (1995), Weng (1998), Pringle (2003), Chopra and Sodhi (2004:
59), Goyal et al. (2006), Van Mieghem (2007), Mayne et al. (2008), Cachon and Terwiesch (2009: 344-
351).
518
519
Pooling server capacity in a queuing system does not affect utilization, as demand is never lost, but might
have to wait longer than desired. The amount of demand fulfilled is independent of the capacity structure
(Cachon and Terwiesch 2009: 346).
520
Weng (1998: 587), Chen and Chen (2003), Plambeck and Taylor (2003), Chen and Ren (2007), Van
Mieghem (2007: 1251), Cachon and Terwiesch (2009: 344).
521
522
Iyer and Jain (2004).
523
Jain (2007).
50
3.5 Sales and Distribution
3.5.1 Product Pooling
Product pooling is the unification of several product designs to a single generic or “universal
design”524 or reducing the number of products or stock-keeping units (“SKU rationaliza-
tion”525) thereby serving demands that were served by their own product variant before with
fewer products.526
The demands for the different products are aggregated to the demand for the universal
design or the reduced number of SKUs, which fluctuates less thanks to risk pooling. Thus
product pooling can reduce demand variability, improve matching of supply and demand,
reduce the demand-supply mismatch cost527, and increase sales and profit or lower invento-
ry for a given target service level.528 Thus, product pooling constitutes a 1/0 – risk pooling
method.
However, a universal design might not provide the desired functionality to consumers
and therefore might not achieve the same total demand as a set of focused designs. It does
not permit price discrimination and may be more expensive than focused designs, because
the components or quality of components of the universal design targeted to many different
uses might not be necessary to some consumers. There might be economies of scale in
production and procurement of a single universal component relative to small quantities of
several components and lower labor costs though.529 The risk pooling benefits of selling a
universal product may compensate its higher cost.530
Product pooling is closely related to postponement, where the differentiation of a uni-
versal product to individual ones is delayed531, standardization, and component commonal-
ity. It may prohibit risk pooling benefits from product substitution.
524
525
Jabbonsky (1994), Lahey (1997), Kulpa (2001), Pamplin (2002), Alfaro and Corbett (2003: 12), Neale et
al. (2003), HTT (2005), Sheffi (2006: 119, 2007), Covino (2008), Harper (2008), Sobel (2008: 172),
Chain Drug Review (2009a, 2009b), Hamstra (2009), MMR (2009), Orgel (2009), Pinto (2009a, 2009b),
Ryan (2009), Thayer (2009).
526
Alfaro and Corbett (2003: 12), Cachon and Terwiesch (2009: 330-336, 467).
527
The demand-supply mismatch cost is the cost of left over inventory (overage cost) and the opportunity
cost of lost sales (underage cost) (Cachon and Terwiesch 2009: 257-263, 433f.).
528
Cachon and Terwiesch (2009: 331), Hamstra (2009), Ryan (2009), Thayer (2009: 23).
529
Cachon and Terwiesch (2009: 335), Hamstra (2009), Ryan (2009), Thayer (2009: 23).
530
Cattani (2000).
531
51
3.5.2 Product Substitution
In product substitution one tries to make customers buy another alternative product, because
the original customer wish is out of stock532 or although it is available (“demand reshape”533).
In manufacturer-driven substitution the manufacturer or supplier makes the decision to
substitute. Typically the manufacturer or supplier substitutes a higher-value or functionality
product, component, or service for a lower-value or functionality one that is not available
(downward substitution).534 Axsäter (2003b: 438) compares this to unidirectional transship-
ments, where transshipments are only allowed in one direction, perhaps because warehouses
differ in their shortage costs. Swaminathan (2001: 130) and Simchi-Levi et al. (2008: 348) relate
substitution to product standardization: With risk pooling through product standardization a
large variety of products may be offered, but only a few kept in inventory. If a product not kept
in stock is ordered, either the product is made or procured or the order filled by downward substi-
tution.
In customer-driven substitution, a customer makes the decision to substitute, because his
original wish is not available.535
If only one of two products is a substitute for the other one, this is called one-way substitu-
tion. In two-way substitution either product substitutes for the other one.536
Substitution allows the manufacturer or retailer to aggregate demand across substitutable
components or products537 (demand pooling538). Demand reshape increases the demand mean
and variability of one product while reducing them for the other product. Substitution and de-
mand reshape reduce total demand variance539 (“demand-pooling effect of substitution”540) and
thus allow to reduce the safety stock for a given customer service level (product availability, fill
rate, or average backlogged demand)541, increase service level without increasing inventory, or a
combination of both542 or increase total profit and raise the customer service level simultaneous-
ly by risk pooling543. Product substitution is a 1/0 – risk pooling method.
532
Swaminathan (2001: 130), Chopra and Meindl (2007: 324ff.), Simchi-Levi et al. (2008: 348).
533
534
Gerchak et al. (1996), Hsu and Bassok (1999), Bassok et al. (1999), Swaminathan (2001: 130), Chopra
and Meindl (2007: 324), Simchi-Levi et al. (2008: 348).
535
Netessine and Rudi (2003), Chopra and Meindl (2007: 324).
536
537
Chopra and Meindl (2007: 324ff.).
538
539
540
Ganesh et al. (2008: 1124).
541
542
Liu and Lee (2007).
543
Weng (1999), Swaminathan (2001: 130), Eynan and Fouque (2003), Simchi-Levi et al. (2008: 348).
52
4 Choosing Suitable Risk Pooling Methods
Findings from our thorough literature review are now merged in a synoptic comparison about
conditions favoring the ten identified risk pooling methods, their advantages, disadvantages,
and basic trade-offs. Based on this synopsis a decision support tool is developed for choosing
appropriate risk pooling methods for a specific situation. Therefore a situational approach is
used.
4.1 Contingency Theory
We adopt a situational approach similar to contingency theory. It claims that there is no “one
best way” as e. g. in Weberian bureaucracy544 to manage a company or make decisions.545 It
depends (is contingent) on internal and external conditions (contingency factors).546 “The best
way to organize depends on the nature of the environment to which the organization must re-
late”547.
Contingency theory is mainly criticized for its positivist548, deterministic, unidirectional,
linear relationships and reduction of complex situations to a limited number of contingency
factors.549
Different from the usual contingency approach we not only use quantitative, empiri-
cal550, but mainly qualitative analysis based on reviewed quantitative analyses, which Do-
naldson (1999) rejects. Kieser and Kubicek (1992: 220ff.), on the other hand, propose to
use empirical correlations not as causalities, but for further interpretations, hypotheses, and
empirical analyses. Höhne (2009: 89) and the references she gives support that contingen-
cy factors are based on both empirical analyses and plausibility assumptions.
We cannot confirm that there is “[o]ne best way for each given situation”551 as our con-
tingency factors are not exhaustive and mutually exclusive and thus several risk pooling
methods may be adequate. However, Donaldson (1996: 118ff.) remarks there are as many
544
Weber (1980: 124ff., 551ff.).
545
Scherer and Beyer (1998: 333f.).
546
Drazin and van de Ven (1985), Miller and Friesen (1980: 268), Scherer and Beyer (1998: 334).
547
Scott (1981: 114).
548
Positivism is “a system recognizing only that which can be scientifically verified or logically proved”
(Soanes and Hawker 2008).
549
Miller and Friesen (1978: 921), Miller (1981: 2f.), Meyer (1993: 1176), Staehle (1994: 49f., 54), Schrey-
ögg (1995: 159ff., 217f., 219), Frese (1992: 193), Scherer and Beyer (1998: 334ff.), Klaas (2002: 99f.),
Höhne (2009: 92f.).
550
Scherer and Beyer (1998: 334), Höhne (2009: 89, 91, 94).
551
Scherer and Beyer (1998: 334).
53
combinations of environment, strategy, and structure (“fits”) as there are different envi-
ronmental situations. Thus, for every situation (combination of contingency factors) there
may be one best risk pooling strategy, i. e. combination and implementation of risk pooling
methods.
Starbuck (1981, 1982: 3), Frese (2000: 488f.), and Schreyögg (2008: 54) consider the
contingency approach obsolete. Others, however, deem it important552, popular, often used,
and useful553, as it is open and flexible554, allows to be combined with other approaches to
advance research555, to make action and design recommendations556, to systematize contin-
gency factors557, and to conduct a practice-oriented analysis558. Therefore it seems ade-
quate for our research.
4.2 Conditions Favoring the Individual Risk Pooling Methods
Table D.3 is based on our thorough literature review and shows which conditions render which
risk pooling methods favorable. The favorable conditions are demand-, lead-time-, transporta-
tion-, service-level-, order-policy-, storage-, product-, process-, company-, and competition-
related. The risk pooling methods (columns) are arranged according to the previous chapter's
structure.
Further research is needed to complete this table and confirm the favorable conditions
for the different risk pooling methods, especially for product pooling, central ordering,
(particularly non-manufacturing) capacity pooling, virtual pooling, and product substitu-
tion/demand reshape. We could only speculate and hypothesize about which risk pooling
methods the stated favorable conditions also apply to. Therefore, we only check marked
them for methods where there is confirmation in the literature and the table is not exhaus-
tive. References and explanations are given in footnotes.
552
Pugh (1981), Donaldson (2001), Daft (2009: 26).
553
For example in management (Hiddemann 2007: 16, Schröder 2008: 68, Müller-Nedebock 2009: 22,
Sommerrock 2009: 88, 93f., Segal-Horn and Faulkner 2010: 157f.), organization (Dzimbiri 2009: 86,
Güttler 2009: 73f., Höhne 2009: 83, 91, 93f., March 2009: 25, Nobre et al. 2009: 31), finance (Pfeifer
2010: 253), information systems (Andres and Zmud 2001, Sugumaran and Arogyaswamy 2003, Khazan-
chi 2005), supply management (Staudinger 2007), business logistics and SCM (Hult et al. 2007, Huang et
al. 2008, Doch 2009: 39ff.), and supply chain risk management (Wagner and Bode 2008) research. The li-
terature database Business Source Complete accessed via EBSCOhost® showed 1,139 results for the
search term “contingency theory” on June 18, 2010.
554
Staudinger (2007: 41), Güttler (2009: 73f.).
555
Sommerrock (2009: 144), Güttler (2009: 73f.).
556
Hiddemann (2007: 16), Schröder (2008: 69), Doch (2009: 39ff.), Pfeifer (2010: 262).
557
Doch (2009: 41).
558
Staudinger (2007: 41), Sommerrock (2009: 94).
54
4.3 The Risk Pooling Methods' Advantages, Disadvantages,

Performance, and Trade-Offs
Table D.4 presents the considered risk pooling methods' possible advantages, disadvantages,
and basic trade-offs in choosing or rejecting them and compares their performance vis-à-vis
other ones.
In summary, the risk pooling methods may, on the one hand, reduce demand and/or lead
time variability, (safety) stock, lead time, as well as procurement, production, transporta-
tion, personnel, and warehouse cost and increase service level (product availability and
fill-rate), utilization, sales, profit, and competitiveness.
On the other hand, they may increase R&D, redesign, component, product, procure-
ment, production, transportation, transaction, and warehousing cost, as well as cycle stock
and cycle time, and decrease customer service, sales, profit, and product functionality. We
only check marked the respective risk pooling method's most important effects that are
confirmed in the studied literature. References are given in footnotes. Explanations are
only given where essential, as we already dealt with the individual risk pooling methods in
detail earlier. These three choices maintain the readability of the table.
Tables D.3 and D.4 can be used to choose suitable risk pooling methods for a specific
company under specific conditions. Scoring methods could be applied or the risk pooling
method with the highest net present value or profit (advantages minus disadvantages ex-
pressed in monetary units) could be chosen. However, the following decision support tool
condenses the most definite distinguishing factors and helps to make a first decision on
possible risk pooling methods elegantly, before carrying out a more thorough cost-benefit
analysis.
55
4.4
/0
*!
1
) 6+
))
,7*! ,2; 3
.
+
+ 3 3
018 *!18

.& . % +
*!18+ 3

"

3

)&

3

*5
)7 $%"
.,#
+ . ,%
,/'( #

+ "

..'(

$%"
3 *5

, 018 *!18

5 - ,. !"#!#
./*! / 6+ , " - *!18 ,,#
+4 3 *5
$%"
5
!# %
.)# + +
% 4 6#
3

3 3

.76+

/#
,)"4
3 ,
.#

4
3
)*! !

5

+
+ 3

+
/2 ,&9

"

3 ) 0

&&#% <

)/#
/& 018 *!18 #

3 3

5 $%"
/ % &'( = 5

3 *!18 5 *5
"

!#
4 6# 3
.2#
+
&7
3

"
$%"
).#

),;

!"#!#
)#

$%"
A Risk Pooling Decision Support Tool

)2:

/7#
3
4
!
+ 5
/) 3 4

3

$%"
3 *!18 :
3
Figure 4.1: Risk Pooling Decision Support Tool

;
< 6

6

&/# <
##:
## #

3
#% #
# #

3 018 0

//"4 2# $%" $
/,
3
4
% &. &,

!
6# 6+

*! "

*!18 *!*!

3 :

/.9 &)#

# " $ 0

56
The Risk Pooling Decision Support Tool (RPDST) depicted by a flow chart in figure 4.1 helps
in choosing a risk pooling strategy to cope with demand and/or lead time uncertainty (1), if
it cannot be reduced more efficiently by other means559.
To facilitate the reader's orientation we introduced numbers in round brackets in this
explanation and the flow chart. We will guide you through the flow chart step by step and
critically revisit it at the end.
(2) If product variety is important, inventory pooling (IP)560, capacity pooling
(CP)561, component commonality (CC)562, postponement (PM)563, product substitution
(PS)564, transshipments (TS)565, virtual pooling (VP)566, centralized ordering (CO), and
order splitting (OS) are a possibility. IP, CP, TS, VP, CO, and OS may also be applied, if
product variety is not important. The first four methods might make more sense, if product
variety is important. CC, PM, and PS are not expedient without product variety.
If product variety is not important, product pooling (PP), IP, CP, TS, VP, CO, and
OS are considered. PP is only meaningful, if product variety is not important. The other
methods may also be applied, if product variety is important.
(3) If product variety is important and the stockout penalty cost high compared to in-
ventory carrying cost (savings), PS567, TS568, VP, CP, CC, PM569, CO570, and OS571 may be
559
In addition to risk pooling it may be reduced by e. g. adding inventory or capacity, having redundant
suppliers, which resembles order splitting, increasing responsiveness (Chopra and Sodhi 2004: 55, 60)
(quick response (Heil 2006)), flexibility, and capability (Chopra and Sodhi 2004: 55, 60), and improving
forecasting (Heil 2006). In LeBlanc et al.'s (2009: 29) model, for instance, PM is favorable unless the
forecast is completely accurate.
560
Pfohl (1994: 142), Schulte (1999: 378), Weng (1999: 80), Vahrenkamp (2000: 221), Alfaro and Corbett
(2003: 28).
561
Jordan and Graves (1995: 577), Cachon and Terwiesch (2009: 344).
562
Srinivasan et al. (1992: 27), Swaminathan (2001: 131), Simchi-Levi et al. (2008: 348), Song and Zhao
(2009: 493).
563
Zinn (1990: 14), Lee and Tang (1997: 52), Pagh and Cooper (1998: 24), Swaminathan and Tayur (1998),
Van Hoek (1998b, 2001: 161, 173), Van Hoek et al. (1998), Aviv and Federgruen (2001a: 514), Ihde
(2001: 36), Matthews and Syed (2004: 34), Graman and Magazine (2006: 1072), Chopra and Meindl
(2007: 363f.), Cachon and Terwiesch (2009: 342, 363). “The potential benefits of postponement in the
presence of product variety are reduced inventory levels and/or improved service levels” (Graman and
Magazine 2006: 1072). “In general, delayed differentiation is an ideal strategy when […] Customers de-
mand many versions, that is, variety is important” (Cachon and Terwiesch 2009: 342).
564
Weng (1999: 80), Swaminathan (2001: 130), Eynan and Fouque (2003), Simchi-Levi et al. (2008: 348).
565
Grahovac and Chakravarty (2001), Wong (2005), Simchi-Levi et al. (2008: 231).
566
Guglielmo (1999), Randall et al. (2002: 56f., 2006: 567), Kroll (2006).
567
Liu and Lee (2007: 1, 41). Only the risk pooling effect is considered, the financial aspect of different
product margins neglected.
568
TS reduce demand and lead time variability by pooling demand and lead times, decrease stockouts, and
thus are sensible, if stockout costs are high (Evers 1997, Needham and Evers 1998, Evers 1999: 133,
2001: 313, Jung et al. 2003, Hu et al. 2005, Wee and Dada 2005: 1519, 1529, Zhao et al. 2005: 545, Lue
2006, Chiou 2008: 428). However, in Jung et al. (2003) the savings from lateral TS decrease with increas-
ing shortage cost from a certain value upwards.
57
applied. PS (costs of persuading a customer to buy a substitute in case of a stockout and

possible customer dissatisfaction because he did not obtain the desired product572), TS
(transshipping costs), VP (drop-shipping or cross-filling costs), CP (large costs to have
flexibility), and CC (common component costs) may only be worthwhile, if stockout costs
are high. PS573, PM574, CO575, and OS576 may also be considered, if the stockout penalty
cost is low. IP is suitable, if the stockout cost is low compared to the inventory (carrying
cost) savings from combining inventories, as it increases the distance between the invento-
ry and the buyer and thus delivery time and therefore might increase stockouts577. As noted
earlier in section 2.4 and table D.3, Dai et al. (2008: 407, 410f.) show when IP by a suppli-
er for two retailers may increase the supplier's and retailers' total profits. We consider IP
for low stockout costs only, as the decision for or against inventory pooling vis-à-vis the
other risk pooling methods is considered.
(4) If (2) is answered in the affirmative, (3) with “High”, and the supplier lead time is
highly variable/uncertain or long, TS578, VP, PM579, CO, and OS580 are considered. TS
569
PM may increase flexibility and responsiveness to customer demand (Zinn and Bowersox 1988: 133,
Herer et al. 2002, Davila and Wouters 2007), increase customer service, decrease stockout costs, and thus
may be sensible, if stockout costs are high.
570
Eppen and Schrage (1981: 51), Schoenmeyr (2005: 5). CO may increase responsiveness to customer
demand, if the allocation of the central order to the delivery warehouses, retailers, or stores can be post-
poned and made more responsively according to more recent demand information (Cachon and Terwiesch
2009: 341), decrease stockout costs, and thus may be sensible, if stockout costs are high.
571
OS can reduce lead time (variability) through lead time pooling and increase inventory availability.
Therefore it can be reasonable, if stockout costs are high (Evers 1999: 122, 132).
572
Swaminathan (2001: 131).
573
PS may be worthwhile in case of low stockout costs, if the sold substitute's profit margin is higher than
that of the original product wish. Besides, even if the original customer wish is available, PS or demand
reshape lowers the total variability of demand and thus inventory costs (Eynan and Fouque 2003, 2005).
574
Besides the risk and stockout cost reduction, PM can lead to economies of scale, synergistic effects, de-
layed increase in costs in the value chain and thus decreased capital commitment, and higher flexibility
(Lee and Tang 1997: 52, Van Hoek 2001: 162, Aviv and Federgruen 2001b: 579, Herer et al. 2002, Kim
and Benjaafar 2002: 16, Ma et al. 2002: 534, Graman and Magazine 2006: 1078, Piontek 2007: 86f.).
Therefore it may be worthwhile, even if stockout costs are not particularly high.
575
PM (Zinn and Bowersox 1988: 125f., Van Hoek 2001: 161, Rabinovich and Evers 2003b: 35, Yang et al.
2005b: 994, Graman and Magazine 2006: 1078, Pishchulov 2008: 13, Mahar and Wright 2009: 3061) and
CO (McGavin et al. 1993: 1094, Cachon and Terwiesch 2009: 341) may lead to a longer delivery time
(from the supplier to the retailer) and more stockouts, and therefore stockout costs should be low. Bucklin
(1965: 27) remarks that speculation, the opposite strategy to postponement, “limits the loss of consumer
goodwill due to stockouts”. CO may also increase stockouts, if it decreases local knowledge (Ganeshan et
al. 2007: 341).
576
Although OS may reduce lead time variability, the split orders are delivered sequentially and thus may
lead to more stockouts than a single order or delivery and stockout costs should be low (Chiang 2001: 73,
cf. Chiang and Chiang 1996).
577
McKinnon (1989: 104), Fawcett et al. (1992: 36ff.), Schulte (1999: 80, 377), Vahrenkamp (2000: 31f.),
Bowersox et al. (1986: 278f.), Evers (1999: 122f.), Ballou (2004b: 573f.), Schmitt et al. (2008a: 14), Ca-
chon and Terwiesch (2009: 336).
578
Evers (1999: 132), Herer et al. (2002), Hu et al. (2005), Wanke and Saliby (2009). However, Evers
(1996) finds the percentage reduction in safety stocks obtained by using non-emergency transshipments
decreases with increasing coefficient of variation of lead time (lead time variability) and average lead
58
and VP by cross-filling may pool demands and lead times, i. e. reduce demand and lead
time variability.581 Long order cycles582 and inflexible production processes583 favor TS.
Long supplier/production lead times support CO.584 For CO the lead time before the distri-
bution center (DC) should be longer than the one after it.585 PM and CO may also reduce
lead time (variability/uncertainty): “Postponement permits firms to be more responsive and
other advantages that are often mentioned are risk-pooling and lead-time uncertainty re-
duction”586. If, for instance, in CO or consolidated distribution product differentiation is
delayed (PM) in the DC, total lead time may be reduced587. Since larger common orders
for multiple locations make more frequent shipments economically possible between the
suppliers and the DC and the DC and the retailers588, consolidated distribution might de-
crease lead time (uncertainty) despite the additional lead time from the DC to the various
retailers. If there is no DC and the retailers merely place joint orders with the suppliers
which are then allocated according to more recent demand information to the retailers just
before the delivery means leave the suppliers, there is no additional lead time between the
DC and the retailers. The retailers' higher demand power through CO and closer collabora-
tion between the retailers and the suppliers might further reduce average lead times and
lead time fluctuations. OS only pools lead times589 and is only worthwhile, if lead time
time at each market (Evers 1996: 126ff.). It increases at a decreasing rate as the proportion of demand at
market i retained at facility i approaches the point where demand at i is spread evenly among all locations
(Evers 1996: 128). By spreading the demands, and therefore the variability, more evenly across locations,
the opportunity to smooth demand fluctuations increases at a decreasing rate (Evers 1996: 128).
579
Transhipments may be seen as a type of time postponement (Rabinovich and Evers 2003b).
580
Kelle and Silver (1990a, 1990b), Pan et al. (1991: 4), Evers (1997, 1999: 122, 132), Kelle and Miller
(2001), Thomas and Tyworth (2006, 2007: 178, 188). According to Thomas and Tyworth (2007: 188),
sequentially releasing a split order according to more recent demand information may reduce demand va-
riability. Chiang and Chiang (1996) and Chiang (2001: 67, 73), however, recommend “splitting an order
into multiple deliveries“ for low demand variability or service level only. The number of stockout possi-
bilities is proportional to the number of deliveries. Thus OS increases stockouts, the more, the more vari-
able demand is. Kelle and Miller (2001: 407) find dual sourcing may decrease stockouts, if there is no
“single, reliable supplier [… and] the variability of the lead-time demand is considerable”, but is only
suitable, if lead time uncertainty cannot be reduced. According to Evers (1999: 123) and Wanke and Sali-
by (2009: 679) splitting a replenishment order into multiple orders only pools lead times, not demand.
Consequently, we recommend OS only for high lead time variability and its lead time pooling, but not its
debatable demand variability reduction effect.
581
Evers (1997, 1999: 121), Zhao et al. (2008).
582
Jönsson and Silver (1987a: 224).
583
Grahovac and Chakravarty (2001: 580).
584
Eppen and Schrage (1981: 51), Schoenmeyr (2005: 5).
585
Cachon and Terwiesch (2009: 341), cf. Kumar and Schwarz (1995).
586
Caux et al. (2006: 3243).
587
Dilts (2005: 43). Personal correspondence with Professor David M. Dilts on July 7, 2009.
588
589
Evers (1999: 122, 132).
59
variability590 and unit values are (unusually) high (in relation to industry standards)591. For
high stockout costs per unit (3), high lead time variability (4), low order quantity and/or
safety stock levels emergency TS perform particularly well compared to OS.592
In case (4) lead times are neither highly variable/uncertain nor long, but rather de-
mand is variable/uncertain and thus demand pooling important, PS593, CP594, CC595, PM596
and CO597 may be appropriate.
590
Kelle and Silver (1990a, 1990b), Pan et al. (1991: 4), Evers (1999: 132), Thomas and Tyworth (2007:
178, 188).
591
Thomas and Tyworth (2007: 178, 188).
592
Evers (1999: 132).
593
Eynan and Fouque (2003, 2005), Chopra and Meindl (2007: 324ff.), Ganesh et al. (2008: 1124).
594
Yu et al. (2008: 1).
595
We will now examine whether CC reduces (the effects of) lead time variability/uncertainty and therefore
should be considered, if it is high:
Benton and Krajewski (1990: 407) show in a computer simulation experiment with 320 observations that
manufacturing “environments with intermediate stocking points and commonality significantly dampen
the effects of lead time uncertainty” (410), “tend[…] to reduce the level of backlogs as well as dampen
their sensitivity to lead time uncertainty” (412), because “commonality causes intermediate inventories to
increase” (413, cf. 410). However, they are more sensitive to vendor quality (“the ability to supply the re-
quested quantity of nondefective parts or raw materials” (404)) and therefore might increase backlogs
(413). Yet, the ambition of risk pooling is to decrease inventory for a given service level or vice versa or a
combination of both. If there are no intermediate stocking points, they are more sensitive to lead time un-
certainty (410). Hence, CC by itself does not reduce “the effects of lead time uncertainty”. Benton and
Krajewski (1990) give no intuitive explanation for the increasing total inventory. Usually common com-
ponents reduce safety and thus total stock, because they are used in several products (Labro 2004: 362).
Eynan and Rosenblatt (1996), Chandra and Kamrani (2004: 175), and Mohebbi and Choobineh (2005:
473) cite Benton and Krajewski (1990) in a truncated manner, aiming at the reduction of lead time uncer-
tainty through CC. Labro (2004) deals with Benton and Krajewski (1990) more accurately.
Although Ma et al. (2002) assume “deterministic replenishment leadtimes” (Ma et al. 2002: 526), they
find “the major benefits of commonality are risk-pooling and lead time uncertainty reduction. These lead
to a safety stock reduction and have been studied extensively by many researchers, including, but not li-
mited to, Baker et al. (1986), Collier (1982), Guerrero (1985), Gerchak et al. (1988), Eynan and Rosen-
blatt (1996), and Grotzinger et al. (1993)” (Ma et al. 2002: 524).
Like most CC research (e. g. McClain et al. 1984, Eynan 1996, Thonemann and Brandeau 2000, Hillier
2002a, 2002b, Thomas et al. 2003, Labro 2004: 359), all these researchers mention demand uncertainty
reduction as a benefit of CC (Baker et al. 1986: 983, Collier 1982: 1300, 1303, Guerrero 1985: 396, Ger-
chak et al. 1988: 755), only Eynan and Rosenblatt (1996) lead time uncertainty reduction in the inade-
quate manner described above.
Under restrictive assumptions, Mohebbi and Choobineh (2005: 476, 481) conclude CC tends to be more
beneficial with both random component procurement order delays and demand uncertainty than with only
one of these uncertainties. They assume, for instance, that the component procurement lead time can
only be longer not shorter than the planned one. Actual uniformly distributed demand for finished
products, however, can be higher or lower than the forecast (476). Costs (481), safety stock and safety
lead times are neglected. Backorders are filled after current demand (475). These researchers do not ex-
plain the causes for their observations. Lead time delays mostly reduce average total inventory per com-
ponent per period (477f.), as they perhaps reduce the time a component remains in storage. This effect in-
creases with increasing demand uncertainty (477), perhaps because a longer-than-planned delivery time
may be offset by lower-than-forecast actual demand. This may be amplified by CC and lead to a higher
service level than without commonality. However, the product with the lowest number of (three) compo-
nents and without commonality shows the lowest inventory per component and the highest service levels
(478f.).
Song and Zhao (2009: 22) even find that in dynamic assemble-to-order systems commonality may not
decrease inventories under the first-in-first-out (FIFO) with identical and under the FIFO and modified
FIFO (MFIFO) rule for dynamically allocating common components to demand with non-identical com-
60
(5) If (2) is “Yes”, (3) “High”, (4) “Yes”, and the fixed cost per order high, TS598, VP,
PM, and CO599 are favorable. If the replenishment order cost is high, one rather resorts to
other cheaper and/or faster possibilities (TS, VP, and PM) than placing a replenishment
order or tries to combine orders by CO to reduce the number of orders and exploit econo-
mies of scale (EOS) and thus lower ordering costs. If it is low600 and there are no transpor-
tation EOS601, OS may be considered. Due to smaller orders OS cannot take advantage of
quantity and transportation discounts and a long-term partnership with a single supplier,
but several suppliers may guarantee competitive pricing and other advantageous terms.602
TS603, VP, and PM may also be applied, if the fixed order cost is low.
(6) In case (2) is “Yes”, (3) “High”, (4) “Yes”, (5) “High”, and the transshipment or
transportation cost is high and there are EOS in ordering, CO604 or PM605 can be
worthwhile. If the transshipment or transportation cost is low and there are no EOS in
ordering, TS606, VP607 or likewise PM608 can be applied. PM may forego EOS:609 Zinn and
ponent replenishment lead times. “The value of commonality tends to decrease as [… the common com-
ponent replenishment lead time] increases under either FIFO or MFIFO”.
If the common component replenishment lead time fluctuates strongly and the common component is not
available, all the products using this component cannot be built. Thus lead time variability/uncertainty of
the common component supplier might be more severe than of a specific component supplier (cf. Benton
and Krajewski 1990: 413, Labro 2004: 363).
All these findings led us to not recommend CC by itself for coping with high lead time variability/un-
certainty and to only consider it for the low lead time variability/uncertainty case.
596
Jain and Paul (2001: 595ff.), Anupindi et al. (2006: 193), Chopra and Meindl (2007: 365), Pishchulov
(2008: 24).
597
If we follow Cachon and Terwiesch (2009: 336, 339, 341f., 349), CO may only pool demands and may
also be applied if lead time (variability) is low. PM or delayed differentiation and CO may increase lead
time and only pool demands during lead time.
598
Herer and Rashit (1999: 525), Daskin (2003), Herer and Tzur (2003: 419).
599
Schwarz (1981: 147), Evers (1995: 5, 1999: 133), Gürbüz et al. (2007: 294).
600
Evers (1999: 132), Thomas and Tyworth (2007: 172, 178, 188).
601
Evers (1999: 122f.), Thomas and Tyworth (2006, 2007), Sajadieh and Eshghi (2009).
602
Evers (1999: 122f.).
603
For TS the fixed order cost is irrelevant (Evers 1999: 132).
604
Eppen and Schrage (1981: 67), Hu et al. (1997, 2005: 34), Gürbüz et al. (2007: 293), Cachon and Ter-
wiesch (2009: 341).
605
Allowing online sales to accumulate and postponing assigning them to a fulfillment site can lead to lower
inventory costs (Mahar and Wright 2009), higher utilization of transportation means, and EOS in order-
ing. PM does not necessarily increase transportation distance, so that transportation costs may be irrele-
vant. Customers may be willing to wait, so that no premium transportation cost is incurred. The product
and process are usually designed, so that the delayed differentiation can be conducted fast (Lee and Tang
1997), the delivery delay is insignificant, and no premium transportation cost is incurred. PM can exploit
EOS (Piontek 2007: 87).
606
Jönsson and Silver (1987a: 224), Robinson (1990), Evers (1999: 132f.), Grahovac and Chakravarty
(2001), Jung et al. (2003), Hu et al. (2005), Anupindi et al. (2006: 191), Kutanoglu (2008: 356f.). Still TS
may save fixed and variable replenishment costs (Herer and Rashit 1999: 525, Herer and Tzur 2003: 419),
if e. g. one location makes a larger replenishment order to transship some of it to other locations especial-
ly in a dynamic deterministic demand environment (Herer and Tzur 2003: 419). Herer and Rashit (1999)
consider a single period random demand environment.
607
Anupindi et al. (2006: 191). In VP small-orders may be drop-shipped more expensively than full truck
loads (Randall et al. 2002, Cachon and Terwiesch 2009: 328f.).
61
Bowersox (1988: 124f.) find that labeling, packaging, and manufacturing PM can lead to
lost EOS. In case of large EOS in the manufacturing and/or logistics processes some
speculation should be applied instead of manufacturing and full PM. Speculation, the op-
posite strategy to PM, rather leads to EOS: “Speculation permits goods to be ordered in
large quantities rather than in small frequent orders. This reduces the costs of sorting and
transportation”610.
(7) If (2) is “Yes”, (3) “High”, (4) “Yes”, (5) “High”, and (6) “Yes”, do you prefer re-
designing your products and/or processes611 to achieve leagility, both agility (respon-
siveness) and leanness (cost minimization), (PM)612 to a less expensive, fast implementa-
ble and changeable solution that keeps inventory closer to customers (CO)613? If yes, you
may implement PM (8), otherwise CO (9). Time and form PM are appropriate, if well-
timed product delivery to and differentiation for the customer are more important than
company-internal cost issues.614 On the one hand, CO may decrease responsiveness be-
cause of a lack of local knowledge about sales615 when ordering. On the other hand, it may
increase responsiveness because of allocating the joint order to the delivery warehouses,
retailers, or stores according to more recent demand information and because aggregate
forecasts usually are more accurate than disaggregate ones616. It is more likely that CO
leads to EOS in ordering than PM. While CO and speculation lead to consolidation of
goods in jointly used transfer and transformation processes and EOS617, PM tends to singu-
larize batches of goods in separate transfer and transformation processes618. To determine
which PM strategy might be most appropriate, you may turn to Zinn and Bowersox (1988),
Cooper (1993), Pagh and Cooper (1998), Yang et al. (2004), and Yeung et al. (2007).
(10) In case (6) is answered in the negative, and the less expensive, fast implementable
and changeable, less cross-functional solution is preferred to redesigning products
608
Bucklin (1965: 27), Zinn and Bowersox (1988: 126), Christopher (1992), Feitzinger and Lee (1997),
Delfmann (1999: 195), Van Hoek (2001: 161).
609
Zinn and Bowersox (1988: 124), Yang et al. (2005b: 994).
610
Bucklin (1965: 27).
611
Products and/or processes may have to be redesigned for PM (Lee and Tang 1997, Van Hoek 1998a, Kim
and Benjaafar 2002, Yang et al. 2005b: 994, Davila and Wouters 2007, Piontek 2007). PM might lead to
higher manufacturing costs (Chopra and Meindl 2007: 365).
612
Feitzinger and Lee (1997), Fliedner and Vokurka (1997), Van Hoek et al. (1999a), Christopher (2000),
Van Hoek (2000a), Christopher and Towill (2001), Herer et al. (2002), Graman and Sanders (2009: 206).
613
Eppen and Schrage (1981), Pishchulov (2008: 22), Cachon and Terwiesch (2009: 349).
614
Alderson (1957), Rabinovich and Evers (2003b: 34).
615
Ganeshan et al. (2007).
616
Cf. footnote 456.
617
Bucklin (1965: 27), Pagh and Cooper (1998: 14), Delfmann (1999: 195), Klaas (2002: 157).
618
Cooper (1983: 71), Bowersox and Closs (1996: 475), Klaas (2002: 157).
62
and/or processes619 and there is no or no significant lead time (uncertainty) reduction

benefit in PM after all, TS and VP are considered. If in addition your supplier, wholesaler,
or warehouse locations have small-order drop-shipping or cross-filling capabilities620
(11), VP (12) may be appropriate. If not, TS (13) may do, especially if the TS cost is lower
than the VP cost. VP may reduce investment in inventory and fulfillment capabilities as
well as overall handling and warehousing costs621 more than TS. However, VP is only re-
commendable, if your company possesses the necessary IS capabilities and is powerful
within the supply chain, because it might lose product margin and control to the drop-
shipping party, which could negatively affect service quality. The drop-shipping party
might bypass the company and directly sell to customers as it possesses all the information
transparency it needs.622 “Virtual inventories provide little benefit in product categories,
such as grocery, where consolidation of orders is necessary but impossible to accomplish
from a few consolidated wholesalers”623.
TS and PM might be used in combination, as emergency TS, an important type of time
PM, support implementing form PM624.
If you do not mind redesigning products and/or processes and a more expensive mid-
to long-term cross-functional implementation and your PM strategy may reduce lead
time uncertainty (10), PM (14) might be the right method to reduce both demand and lead
time uncertainty. TS can increase leagility like PM625, but perhaps more inexpensively and
faster.
Returning to (5), if the fixed order cost is low, and (15) both demand uncertainty re-
duction and lead time (uncertainty) reduction and fewer order placements are pre-
ferred to only lead time uncertainty reduction, but avoiding premium transportation
and in-transit inventory costs, TS626, VP627, and PM628 come into question and the al-
619
Lee and Tang (1997: 40, 1998), Van Hoek (1998a), Kapuscinski and Tayur (1999), Jain and Paul (2001),
Swaminathan (2001: 129f.), Herer et al. (2002), Kim and Benjaafar (2002: 12), Yang et al. (2005b: 994),
Davila and Wouters (2007: 2245), Shao and Ji (2008), Simchi-Levi et al. (2008: 345f.).
620
Randall et al. (2002: 56), Cachon and Terwiesch (2009: 328f.).
621
622
623
624
Rabinovich and Evers (2003b: 41f.).
625
Herer et al. (2002: 201), Zhao et al. (2008).
626
Tagaras (1989), Tagaras and Cohen (1992), Evers (1997, 1999), Hong-Minh et al. (2000), Herer et al.
(2002), Çömez et al. (2006).
627
628
Jain and Paul (2001: 595ff.), Yang et al. (2004: 1054), Dilts (2005: 43), Anupindi et al. (2006: 193), Caux
et al. (2006: 3243), Chopra and Meindl (2007: 365), Pishchulov (2008: 24).
63
ready explained decision process (10) through (14) starts. If (15) is negated, OS629 (16)
may be implemented.
Moving upwards again, in case (3) is low and (17) inventory reduction is more impor-
tant than agility (customer responsiveness), IP630 and PS631 are considered. If (17) is ne-
gated, PM, PS, CO, and OS come into consideration. PM improves agility632 and flexibili-
ty633 and enables product variety634, but may not decrease inventory as much as IP or PS. PS
performs better than IP in terms of location accessibility and lost sales (customer responsive-
ness)635, but may not decrease inventory as much as IP. With PS the customer may get a
product right away, but not the originally desired one. CO performs worse than IP in terms of
inventory reduction, but keeps inventory near customers636. The allocation of the central or-
der to the delivery warehouses, retailers, or stores can be postponed and made more respon-
sively according to more recent demand information. Reducing lead time uncertainty OS
may lower (safety) stock for a given service level637, but not as much as IP, as the former
only pools lead times, not demands638. Simultaneously splitting an order among suppliers
can reduce cycle stock due to successive deliveries of smaller split orders639, but not the sum
of in-transit and cycle stock in the system640. OS may be more responsive than IP, because it
reduces lead time variability and keeps inventory close to customers. However, OS may in-
crease stockouts, if demand variability is high due to successive deliveries.641
In case (17) is confirmed, and (18) the fast implementable, changeable, less expensive
(especially in terms of transportation cost) solution with higher location accessibility and
629
Evers (1999: 132), Thomas and Tyworth (2007: 172, 178, 188).
630
IP may reduce inventory more than PS, but creates distance between inventory and customers and there-
fore may reduce product availability and profits (Bowersox et al. 1986: 278f., McKinnon 1989: 104,
Fawcett et al. 1992: 36ff., Evers 1999: 122f., Schulte 1999: 80, 377, Vahrenkamp 2000: 31f., Ballou
2004b: 573f., Schmitt et al. 2008a: 14, Cachon and Terwiesch 2009: 336).
631
PS might be seen as not as responsive as PM because the customer does not receive his original wish and
is persuaded to buy a substitute, which may cause some customer dissatisfaction, but might reduce inven-
tory more than PM.
632
Zinn and Bowersox (1988), Herer et al. (2002), Davila and Wouters (2007).
633
Lee and Tang (1997: 52), Feitzinger and Lee (1997: 117ff.), Van Hoek (2001: 162), Ma et al. (2002:
534), Graman and Magazine (2006: 1078), Piontek (2007: 86f.).
634
635
636
Eppen and Schrage (1981), Cachon and Terwiesch (2009: 349), Pishchulov (2008: 22).
637
Evers (1999: 122, 132), Thomas and Tyworth (2006: 245).
638
Evers (1999: 122, 132).
639
Moinzadeh and Nahmias (1988), Pan and Liao (1989), Ramasesh (1990), Sculli and Shum (1990), Pan et
al. (1991), Ramasesh et al. (1991), Hong and Hayya (1992), Zhou and Lau (1992), Lau and Zhao (1993),
Ramasesh et al. (1993), Chiang and Benton (1994), Lau and Zhao (1994), Gupta and Kini (1995), Chiang
and Chiang (1996), Hill (1996), Ganeshan et al. (1999), Chiang (2001), Ghodsypour and O'Brien (2001),
Ryu and Lee (2003), Thomas and Tyworth (2006: 245).
640
641
Chiang and Chiang (1996) and Chiang (2001: 67, 73).
64
fewer lost sales is preferred to product functionality and inventory reduction, PS (19)
may be suitable. If (18) is negated, IP (20) might be the right method. In contrast to PS, IP
entails aggregate and thus more accurate forecasts642. Individual demand forecasts for con-
trolling the previously separate inventories are replaced by a more accurate aggregate fore-
cast for managing the consolidated inventory. Of course, PS can only be applied, if prod-
ucts or parts are substitutable, which the preceding affirmation of question (2) suggested.
If (17) is negated and (21) lead times are highly variable/uncertain or long, PM, CO,
and OS are considered (cf. (4)). If then (22) the fixed order cost is high (CO and PM may
be appropriate (cf. (5))) and (23) a perhaps cheaper solution, that keeps inventory closer
to customers, is preferred to redesigning products and/or processes to offer customized
products more responsively and there are EOS in ordering643, CO (24) may be chosen. If
(23) is negated, PM (25) might be appropriate.
If (22) is low (OS and PM may be appropriate (cf. (5))) and product and/or process
redesign to reduce demand uncertainty and perhaps lead time (uncertainty) (cf. (4)) is
preferred to only lead time pooling and keeping inventory close to customers, but in-
curring higher ordering costs (26), then PM (25) might be suitable. Otherwise, OS (27)
may be worthwhile.
If (21) is negated and rather demand pooling is important, PM, PS, and CO are scruti-
nized (cf. (4)). If then (28) the less expensive, fast implementable, and changeable solu-
tion is preferred to giving the customer always the desired product (not a substitute), but
in a longer delivery time and with product and/or process redesign, PS (19) may be
appropriate, otherwise PM and CO are considered. If in this latter case (29) the cheaper
solution that keeps inventory closer to customers is preferred to product and/or
process redesign and leagility and there are EOS in ordering, then CO (24) may be ap-
propriate, otherwise PM (14) (cf. (23) to (25)).
If the importance of product variety is denied in (2) and lead times are highly varia-
ble/uncertain or long (30), TS, VP, OS, and CO are considered (cf. (4)). In case (30) is ne-
gated and rather demand uncertainty reduction is important, PP644, IP645, CP, and CO646 (cf.
(4)) are considered as they only pool demands and thus reduce demand variability, but not
lead time variability. According to Dilts (2005) CO may also reduce lead times (cf. (4)).
642
Cf. footnote 456.
643
CO may utilize EOS in procurement (Eppen and Schrage 1981: 67, Hu et al. 1997, 2005: 34, Gürbüz et
al. 2007: 293, Cachon and Terwiesch 2009: 341).
644
645
Evers (1996: 115, 1997: 57, 60, 1999: 121).
646
Cf. Cachon and Terwiesch (2009: 336, 339, 341f., 349).
65
If (30) is affirmed, (31) the fixed order cost is high (CO, TS, and VP are considered (cf.
(5))), (32) the transshipment or transportation cost is high and there are EOS in order-
ing, CO (33) may be reasonable (cf. (6)). If (32) is negated and (34) your supplier, wholesa-
ler, or warehouse locations have small-order drop-shipping or cross-filling capabilities647
(cf. (11)), VP (35) may be appropriate, especially if it is less expensive than TS. If not, TS
(36) may do (cf. (11) to (13)).
If (31) is low (OS, TS, and VP are considered (cf. (5))) and (37) both demand uncer-
tainty and lead time (uncertainty) reduction, fewer stockouts, and order placements are
preferred to only lead time pooling, but avoiding premium transportation and in-transit
inventory costs of stock transfers648, then TS and VP are considered further and the already
explained decision process (34) to (36) is triggered again. If (37) is answered in the negative,
OS (38) may be economically sensible.
If (30) is negated and (39) accommodating demand uncertainty649 is preferred to
reducing inventory and system utilization (arrival divided by service rate) is high, CP and
CO are considered. “While the relative benefit of inventory pooling tends to diminish with
utilization, the relative benefit of capacity pooling tends to increase with utilization” in a
production-inventory system with endogenous supply lead times.650 CP or manufacturing
flexibility is more valuable with approximately equal total capacity and expected demand.651
CO does not reduce inventory as much as IP, but keeps it near customers.652 The postponed
more responsive allocation of the central order to the delivery warehouses, retailers, or stores
according to more recent demand information enables better matching of supply and de-
mand653 (also cf. (17)). If afterwards EOS in ordering and a preference for a possibly less
expensive, fast implementable and changeable solution are affirmed in (40), CO (33) may
be appropriate. If (40) is negated and (41) CP/(manufacturing) flexibility is cheaper than
capacity, CP (42) might be applied654. Goyal and Netessine (2005) may help to choose a
type of manufacturing flexibility (volume and/or product flexibility)655. If capacity is cheap-
647
648
Evers (1997: 72, 1999: 123f.).
649
650
Benjaafar et al. (2005: 565).
651
652
Eppen and Schrage (1981: 61), Pishchulov (2008: 22), Cachon and Terwiesch (2009: 349).
653
654
655
Volume flexibility is “[t]he ability to adjust the output volume of a product without incurring large costs“,
product flexibility “the ability to manufacture diﬀerent products and to efficiently shift production from
products with low demand to products with high demand” (Goyal and Netessine 2005: 2). For a monopol-
ist, product flexibility reduces uncertainty in demand for individual products more and in aggregate de-
66
er than CP/flexibity, adding capacity (43) might be a better choice.656 CP may be more
appropriate for manufacturing, CO for trading companies (cf. (48)).
If (39) is negated, PP and IP (cf. (17)) are considered, as inventory reduction is given
priority. If then (44) product availability is preferred to product functionality, PP (45)
may lead to a higher product availability than IP (46): It does not increase the distance be-
tween products and customers and may enable EOS in production and procurement of a sin-
gle universal component or product, but might degrade product functionality, not achieve the
same total demand as a set of focused designs, and eliminate some brand/price segmentation
opportunities.657
Returning to (4), if lead times are not highly variable/uncertain or long, rather demand
pooling is important (PS, CP, CC, PM, and CO are considered), and (47) your products share
components or qualities to satisfy the same or similar needs or you are willing to create these
commonalities, then PS, CP, CC, PM, and CO may be applied.
Substitute goods satisfy the same or similar needs and therefore are seen as an alternative
or replacement by consumers. The reason for this replacement relationship is the functional
replaceability between two goods, if they correspond in price, quality, and utility or perfor-
mance so that they are able to meet the consumer's need. This thus enables substitution struc-
turally.658
CP is also based on commonalities: Common design, assembly processing, training in
common methods, suppliers capable of coping with common design and processing and
reacting to volume and complexity are needed.659 CC designs products that share compo-
nents.660
PM advantages partly arise from CC.661 CO does not require (component) commonality.
However, the centrally ordered products usually are needed by several requisitioners (deli-
very warehouses, stores, or departments). Therefore one could say the products are common
to two or more locations.
mand less than volume flexibility. Moreover, product flexibility copes better with substitutable and worse
with complementary products than volume flexibility. Using both volume and product flexibility can lead
to diseconomies of scope (Goyal and Netessine 2005: 1).
656
657
658
Wildemann (2008: 72).
659
Mayne et al. (2008).
660
Chopra and Meindl (2007: 326ff.).
661
Zinn and Bowersox (1988: 124f.), Lee (1996, 1998), Van Hoek (1998b, 2001: 173), Van Hoek et al.
(1998), Feitzinger and Lee (1997), Pagh and Cooper (1998), Swaminathan and Tayur (1998), Aviv and
Federgruen (2001a, 2001b: 579), Swaminathan (2001), Yang et al. (2005b: 993), Simchi-Levi et al.
(2008).
67
PM and CP perform better with similar demand. For PM that means that demand is of
comparable size662 or that the standard deviations of demand for items in the product line
are similar663, for CP approximately equal total capacity and expected demand664 and simi-
lar demand variability665. Capacity sharing among independent companies is more benefi-
cial, if they are similar in their characteristics (work content, demand rates, or delay
costs).666 PM may reduce EOS667 or increase the cycle/delivery lead time668.
If then (48) a less expensive, fast implementable and changeable solution is preferred
to product and/or process redesign and leagility and you run a trading or retail com-
pany rather than a pure manufacturing company, then PS and CO may be applied, other-
wise CP, CC, and PM.
PS can be implemented faster than CP, CC, and PM. It may only incur the costs of per-
suading a customer to buy a substitute669 and maybe cause customer dissatisfaction with
the substitute purchase670.
CO may be less responsive, but also less expensive, implemented within a shorter time
frame, and involve less departments (mainly purchasing and sales) than CP, CC, and PM.
It does not require a product redesign either. CP671, CC672, and PM673 may necessitate a
costly product and/or process redesign and therefore also involve at least manufacturing
and R&D in addition.
As products are only delivered after a customer order is received, in general PM leads to
a centralized logistics system674 and may increase the distance between inventory and cus-
tomers relative to CO675 and even more so compared to PS. With CP a product is produced,
662
Chopra and Meindl (2007: 363f.).
663
Zinn (1990: 14), Alfaro and Corbett (2003: 18f., 24), Cachon and Terwiesch (2009: 342).
664
Jordan and Graves (1995: 583), Cachon and Terwiesch (2009: 348).
665
Iyer and Jain (2004), Jain (2007).
666
Yu et al. (2008: 26).
667
Zinn and Bowersox (1988: 124), Yang et al. (2005b: 994).
668
Zinn and Bowersox (1988: 126), Van Hoek (2001: 161), Rabinovich and Evers (2003b: 35), Yang et al.
(2005b: 994), Graman and Magazine (2006: 1078), Pishchulov (2008: 13), Mahar and Wright (2009:
3061).
669
670
671
Jordan and Graves (1995), Mayne et al. (2008), Cachon and Terwiesch (2009: 349).
672
Bagchi and Gutierrez (1992: 817), Swaminathan (2001: 129, 131), Ma et al. (2002: 536), Simchi-Levi et
al. (2008: 345f., 348).
673
Lee and Tang (1997: 40, 1998), Van Hoek (1998a), Kapuscinski and Tayur (1999), Jain and Paul (2001),
Swaminathan (2001: 129f.), Kim and Benjaafar (2002: 12), Davila and Wouters (2007: 2245), Yang et al.
(2005b: 993f.), Piontek (2007), Shao and Ji (2008), Simchi-Levi et al. (2008: 346).
674
Shapiro and Heskett (1985: 53), Zinn and Bowersox (1988: 126), Pfohl (1994: 145), Pagh and Cooper
(1998: 14), Van Hoek (1998a: 95).
675
Eppen and Schrage (1981), Pishchulov (2008: 22), Cachon and Terwiesch (2009: 349).
68
stored, or transported, or a customer served where there is capacity, so that the distance
between inventory and customers may increase.
PM research “is highly conceptual”676. Applications are mostly limited to manufactur-
ing677, e. g. in Zinn and Bowersox (1988: 121), Lee et al. (1993), Dröge et al. (1995), Feit-
zinger and Lee (1997), Garg and Tang (1997), Lee and Tang (1998: 172), Van Hoek
(1997), Dominguez and Lashkari (2004), Caux et al. (2006), Chopra and Meindl (2007:
365), Davila and Wouters (2007), Harrison and Skipworth (2008: 193), Kumar et al.
(2009), and Kumar and Wilson (2009). Bailey and Rabinovich (2005) exceptionally apply
PM in internet book retailing. “Currently, relatively simple postponement applications are
practiced in wholesale and logistics services as supplementary services. More complex,
high value adding manufacturing activities are still the primary domain of manufacturers
and these are not often outsourced to logistics service providers”678.
Our survey results in appendix A support that CC679, PM, and CP680 are predominantly
applied in manufacturing, CO or consolidated distribution681 and PS682 rather in trade, so
that their implementation might be more promising here. PS is more efficient than CC.683
In case (48) is answered in the affirmative (PS and CO are considered) and
(49) matching supply and demand and product functionality are preferred to a less
costly, fast implementable and changeable solution with higher (substitute) product
availability, but possible customer dissatisfaction because of substitution and there are
EOS in ordering, then CO (33) may be appropriate, otherwise PS (50). PS offers the cus-
tomer a product right away, however, not the originally desired one but a substitute, which
may lead to customer dissatisfaction. CO gives the customer the desired product maybe in
a longer lead time or the sale is lost. PS can be implemented and changed faster and less
expensively than CO. The latter may benefit from EOS in procurement though (cf. (23)).
676
Yang et al. (2005b: 994).
677
Boone et al. (2007: 594).
678
Van Hoek and Van Dierdonck (2000: 205).
679
Gerchak and Henig (1989), McDermott and Stock (1994), Tsubone et al. (1994), Sheu and Wacker
(1997), Swaminathan and Tayur (1998), Swaminathan (2001), Hillier (2002b), Ma et al. (2002), Labro
(2004), Mohebbi and Choobineh (2005), Chew et al. (2006), Simchi-Levi et al. (2008: 345-348), Nonås
(2009), Wazed et al. (2008, 2009: 76).
680
Upton (1994, 1995), Jordan and Graves (1995), Weng (1998), Plambeck and Taylor (2003), Pringle
(2003), Iyer and Jain (2004), Benjaafar et al. (2005), Jain (2007), Goyal et al. (2006), Van Mieghem
(2007), Mayne et al. (2008).
681
Eppen and Schrage (1981), Cachon and Terwiesch (2009: 336-341).
682
Anupindi et al. (1998), Smith and Agrawal (2000), Mahajan and Van Ryzin (2001a, 2001b), Rajaram and
Tang (2001), Eynan and Fouque (2003, 2005), Zhao and Atkins (2009).
683
Eynan and Fouque (2005).
69
If (48) is negated (CP, CC, and PM are considered), (51) the costs to have CP or
(manufacturing) flexibility are lower than the costs for CC (common components, product
redesign, and possible cannibalization) and PM (product and/or process redesign or rese-
quencing), then the already explained decision process (41) through (43) has to be com-
pleted.
If (51) is answered in the negative (CC and PM are considered) and the product684 (cf.
(47)) and (52) (manufacturing) process is modular685, “the firm maximizes its perfor-
mance by adopting”686 PM (53): It “will help to maximize effective forecast accuracy and
minimize inventory costs”687. In case the product is modular688 (cf. (47)) but the process
is not689, then differentiation cannot be delayed, but CC (54) “is the most effective ap-
proach”690. Simchi-Levi et al. (2008: 348) express this more cautiously: CC “is likely to be
effective”. However, we follow the primary source Swaminathan (2001: 132). To deter-
mine the optimal level of CC you may refer to Boysen and Scholl (2009).
Nevertheless, if the process (and the product) is modular, CC and PM may both be ap-
plied. To choose between them the CC691 cost (product redesign692 and possible cannibali-
zation of higher priced components693) has to be traded off against the PM cost (higher
production, transportation, and process and/or product redesign cost (cf. (7))). This be-
684
Feitzinger and Lee (1997), Lee (1998), Van Hoek et al. (1998), Swaminathan (2001: 131), Van Hoek
(1998a, 1998b, 2001: 173), Chiou et al. (2002: 113, 122), Yang et al. (2004: 1053), Simchi-Levi et al.
(2008: 348), ElMaraghy and Mahmoudi (2009: 483).
685
Feitzinger and Lee (1997), Van Hoek et al. (1998), Swaminathan (2001: 131), Van Hoek (2001: 173),
Simchi-Levi et al. (2008: 348). A modular product, e. g. a personal computer as opposed to a shirt, is as-
sembled from a variety of modules such that there is a number of options the customer is interested in for
each module. This is important for concurrent and parallel processing (Swaminathan 2001: 128, Simchi-
Levi et al. 2008: 345).
A modular (manufacturing) process, e. g. semiconductor wafer fabrication as opposed to oil refining, con-
sists of discrete operations, so that inventory can be stored in semi-finished form between operations.
Products are differentiated by undergoing a different subset of operations during the manufacturing
process. Modular products are not necessarily made in modular processes (Swaminathan 2001: 128, Sim-
chi-Levi et al. 2008: 345).
686
687
Simchi-Levi et al. (2008: 348).
688
Weng (1999), Swaminathan (2001: 131), Yang et al. (2005b: 993), Simchi-Levi et al. (2008: 348).
689
Swaminathan (2001: 131), Simchi-Levi et al. (2008: 348).
690
691
The common component may be more expensive than the dedicated ones that it replaces (Hillier 2002a,
2002b, Van Mieghem 2004: 419, Zhou and Grubbström 2004, Mohebbi and Choobineh 2005: 473). Still
component commonality may reduce total cost (Van Mieghem 2004: 419, Hillier 2002a, 2002b).
692
Sometimes, it is necessary to redesign products and/or processes to achieve commonality (Bagchi and
Gutierrez 1992: 817, Swaminathan 2001: 129, 131, Ma et al. 2002: 536, Simchi-Levi et al. 2008: 345,
348), which may result in a smaller and more economical set of components though (Whybark 1989).
693
Excessive part commonality can reduce product differentiation, so that less expensive customization
options might cannibalize sales of more expensive parts (Ulrich and Ellison 1999, Kim and Chhajed
2000, 2001: 219, Desai et al. 2001, Krishnan and Gupta 2001, Ramdas and Sawhney 2001, Simpson et al.
2001, Swaminathan 2001: 131, Heese and Swaminathan 2006, Simchi-Levi et al. 2008: 348).
70
comes important if not form, but place and time PM are considered. CC and PM often are
connected and can be applied together. Maybe applying both of them may be more eco-
nomical than applying only one. PM implementation may require CC694 (“postponement
through standardization”695). According to Caux et al. (2006: 3244), PM in the form of
delayed differentiation can be implemented by “process restructuring, component commo-
nality and product design”.
If (47) is negated, then CO and CP are considered, because they do not require CC, es-
pecially if transportation or storage capacity is pooled. Yet, we consider CP related to in-
ventories under IP. If then (55) the costs of CO are lower than both the ones of CP and
adding capacity and there are EOS in ordering, then CO (33) may be effective. Other-
wise CP and adding capacity are considered and the already explained decision process
(41) to (43) still has to be followed.
Frequently the risk pooling methods cannot be separated mutually exclusively, because
they are favorable under similar conditions as tables D.3 and D.4 already showed. If the de-
cisions in the RPDST are not selective and disjunctive, the respective risk pooling methods
are considered for both choices. The RPDST only shows tendencies and enables general
statements about the advantageousness of risk pooling design options, since the specific size
of the costs and benefits are unknown and the complex interdependencies are simplified. The
advantageousness of specific risk pooling methods in a specific situation for a specific
company depends on specific cost parameters and their interaction or ratios.696
In a particular company case risk pooling design recommendations might differ from
this RPDST. The identified conditions and recommendations for risk pooling methods de-
pend on the selected reference framework. Therefore a study with the same topic might
attain a different classification and thus different results. The framework gives general rec-
ommendations about the advantageousness of risk pooling methods. Afterwards a compa-
ny should weigh the costs and benefits of the different suitable options expressed in mone-
tary units to reach a decision about which risk pooling strategy to implement. Expenses for
implementing the risk pooling strategy might exceed the achievable savings.697
Some benefits besides decreased (inventory) cost such as increased flexibility, respon-
siveness, and customer service level are hard to quantify. Resequencing due to PM might
lower inventory levels, but increase the inventory value per unit. However, if customiza-
694
695
Ma et al. (2002: 534).
696
Cf. e. g. Wanke (2009) on IP.
697
Cf. Simchi-Levi et al. (2008: 349).
71
tion is postponed, the generic products may have a lower value than the customized ones.
Tariffs and duties for undifferentiated products might be lower. Still products and
processes might be more expensive under some risk pooling methods.698
Implementing risk pooling might lead to positive or negative effects not studied in detail
in this RPDST, such as human resistance or negative impacts on the whole supply chain.
It might be appropriate to apply risk pooling methods just to certain products (e. g.
the ones with uncertain demand).699
Combining several risk pooling methods might improve performance.700 Using, for
instance, time PM (e. g. emergency TS and inventory centralization), form PM (product
customization), and enterprise-wide information systems together improved “enterprise-
wide inventory management performance” more than using each of these methods sepa-
rately in an empirical model based on 216 large and established U.S. manufacturing
firms.701 (Logistics) PM might go hand in hand with IP (inventory centralization) or CO
(centralized distribution).702 A hybrid strategy of pooling demand by centralizing invento-
ries (IP) but splitting supply by multisourcing (OS) can be worthwhile.703 Substitutable
products support the implementation of variety PM.704 Demand substitution furthermore
facilitates flexible manufacturing without competition and may facilitate its adoption with
competition.705 CP may exploit CC.706
Kutanoglu (2008) considers both emergency lateral shipments from other local facilities
and direct shipments (VP in the sense of Netessine and Rudi 2006) from one central ware-
house of service parts to repair high-technology hardware systems such as computers and
telecommunication devices in case of a stockout at a local facility within the time window
necessary for the target service level. Alfredsson and Verrijdt (1999: 1430) find conjoint
use of lateral TS and direct deliveries can reduce costs significantly, particularly for slow
moving parts. Investing only in direct delivery flexibility may increase costs dramatically,
unless pipeline flexibility is applied. Pipeline flexibility means that a customer backorder is
satisfied by the part that arrives first either via normal replenishment or via direct delivery.
Similarly, Wong et al. (2007b: 1056) remark emergency direct deliveries from the central
698
Simchi-Levi et al. (2008: 349).
699
Cf. Chopra and Meindl (2007: 366).
700
Cf. e. g. Hong-Minh et al. (2000).
701
Rabinovich and Evers (2003b: 38, 42f.).
702
Shapiro (1984), Zinn and Bowersox (1988), Van Hoek (1998a, 2001), Nair (2005: 451).
703
Benjaafar et al. (2004a: 1442).
704
Heskett and Signorelli (1984), Kopczak and Lee (1994), Goyal and Netessine (2005: 22).
705
Goyal and Netessine (2006: 1).
706
Mayne et al. (2008).
72
warehouse and the plant in case of stockouts in a two-echelon system are only worthwhile,
if lateral TS between local warehouses are not possible.
The RPDST neglects the degree to which the different favorable risk pooling methods
should be applied and their opposite methods. Often two opposing methods are applied in
combination to a certain degree (e. g. centralization and decentralization, consolidation and
singularization, postponement and speculation). Speculative inventories, for instance,
should be held in the logistics channel wherever their costs do not exceed the savings
achievable by postponement.707 Furthermore, the certain part of the product line or the ge-
neric product can be made less expensively to stock, the uncertain products or product va-
riants to order.
An empirical verification of the derived results pertaining to the design recommenda-
tions appears worthwhile. The theoretical analysis revealed fundamental interdependencies
of risk pooling methods and facilitates the practical risk pooling design. Possible devia-
tions from business practice can be disclosed by empiricism. The analytic process can be
adapted accordingly. Therefore, we will apply the RPDST to a German paper wholesaler to
determine risk pooling methods that may reduce the demand and lead time uncertainty it
faces.
707
Bucklin (1965: 28).
73
5 Applying Risk Pooling at Papierco
In this chapter we will first introduce Papierco to you and some problems it faces. Afterwards
we will determine risk pooling methods with the help of the RPDST developed in the previous
chapter to solve these problems and consider their application at Papierco in more detail. Final-
ly we will summarize our results.
5.1 Papierco
A German paper wholesale company (to protect confidentiality we call it Papierco in the fol-
lowing) pursues the business strategy of a decentralized, differentiated full-line distributor and
system provider of paper, printing accessories, as well as screenprinting and advertising sys-
tems, which precludes central warehouses and minimal storage. This corporate strategy seems
to be Papierco's competitive advantage, as its competitors cannot offer such a ubiquitous and
fast delivery service as well as broad assortment. The decentralized warehouse network also
constitutes a competitive advantage in the light of increasing fuel costs, autobahn toll, climate
change, environmental legislation, and increased traffic volume. Papierco delivers its products
from its warehouses with its own truck fleet or has them drop-shipped from paper plants to its
customers.
According to its management, Papierco is a full-line wholesaler, as its 32,739 customers
(mainly printing offices) need all kinds of different product specifications mostly within 24
hours. If a product with a low turnover was eliminated from the assortment, some custom-
ers might not buy some other products either anymore due to product interdependencies.
The alliance Papierco of several legally independent paper wholesalers and exchange of
items between 21 regional warehouses in case of a stockout (emergency transshipments)
allow the individual companies and locations to offer a larger product assortment than they
could store on their own. Not every of the 7,000 product specifications Papierco offers is
stored at every location. Products with a low turnover share are only stocked at few loca-
tions due to cost considerations.708 This is called “selective stocking”709, selective stock
keeping710, or “specialization”711. The idea behind selective stock keeping is lowering in-
ventory holding costs by treating products differently without impairing the service level
708
Pfohl (2004a: 122f.), Simchi-Levi et al. (2008: 233).
709
Ross (1996: 310f.).
710
Pfohl (2004a: 118ff.).
711
Anupindi et al. (2006: 191f.).
74
considerably.712 Transshipments between Papierco's locations are necessary, so that every

location can offer the complete assortment. The shipping decision is postponed until a cus-
tomer order is received. Every warehouse thus functions as both a regional and central
warehouse and can access products in every other warehouse as if all warehouses were one
single warehouse (virtual inventory or warehouse). Hence Papierco takes advantage of
inventory pooling, but mitigates some of the disadvantages of a central warehouse
through the use of transshipments and information technology.713
5.2 Problems Papierco Faces
5.2.1 Fierce Competition in German Paper Wholesale
Paper wholesale in Germany is characterized by strong competition especially in office, illu-

stration printing, and offset paper: Sales stagnate, sales prices decrease, raw material, energy,
and logistics costs increase. A lot of customers go bankrupt and there is an ongoing concentra-
tion process in the printing industry.714 Paper manufacturers, the paper wholesale's suppliers,
suffer from bad profitability due to increasing costs, overcapacities, and price pressure715, as
well as competition pressure from Asia and Eastern Europe716, which they seek to improve by
mergers and markups. The current international economic downturn intensifies the cost and
competition pressure in all industrial sectors. Insufficient capacity utilization is a strain on the
paper wholesale, its customers, and suppliers. Increasing profit margin pressure is believed to
promote the consolidation tendency on all levels of the value chain.717 Thus paper wholesale
faces strong competition and economically stricken customers and suppliers, whose negotia-
tion power increases because of mergers. Competitiveness and economic success can be se-
cured on this market by efficient supply, storage, and distribution planning enabled by risk
pooling, particularly since logistics incurs the highest costs at Papierco.
5.2.2 Supplier Lead Time Uncertainty
The lead time of paper manufacturers fluctuates strongly, ranges from 14 to 154 days, and is
on average 28.49 days. It fluctuates subject to price increases, drop-shipping sales, (interna-
tional) demand, and production setup times. Supplier lead times are entered into the enterprise
712
Pfohl (2004a: 122).
713
Cf. Sussams (1986: 8f.), Simchi-Levi et al. (2008: 233).
714
BVdDP (2009).
715
Kibat (2007).
716
FPT (2007b: 7).
717
BVdDP (2009).
75
resource planning (ERP) system by a central master data maintenance in three German Papier-
co locations. The actual lead time of each supplier should rather be recorded for every delivery
to improve the purchasing policy.
Because of long and uncertain lead times for certain products paper merchants some-
times place so called “phantom orders” for product quantities they do not actually need as
a precaution, which intensifies the bullwhip effect718 and leads to even longer and more
variable lead times. This once resulted in all orders having to be cancelled and only enter-
ing real orders (with a corresponding real demand) again afterwards. Thus lead times nor-
malized again.
5.2.3 Customer Demand Uncertainty
Total Sales (t)
320,000 2009
2008
290,000 2007
260,000
230,000
200,000
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Months
Figure 5.1: Total Fine Paper Sales in Tons per Month in Germany (BVdDP 2010a: 8)
718
Lee et al. (1997a: 98).
76
115000
110000
105000
100000
Total Sales (t)
95000
2007
90000 2008
2009
85000
80000
75000
70000
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Months
Figure 5.2: Papierco's Total Fine Paper Sales to Customers in Germany in Tons per Month
Total demand for fine paper in Germany is seasonal: It is higher in January, February, and
March, lower from April till July, and increases up to November in order to decrease again in
December as figures 5.1 and 5.2 show. Total customer demand depends on the overall eco-
nomic development, i. e. on the available financial means for e. g. new products and advertis-
ing budgets.
Warehouse sales made up 42 %, drop-shipping sales 58 % of all sales in 2007 and
2008,719 respectively 43 % and 57 % in 2009720. Warehouse sales of individual products
per week per warehouse are sporadic without a trend or seasonal pattern as figure 5.3
shows exemplarily.
719
BVdDP (2009).
720
BVdDP (2010b).
77
Figure 5.3: Papierco's 2007 Warehouse Sales to Customers per Week of Seven Standard
Paper Products from the Hemmingen Warehouse
78
“When demand for items is intermittent, because of low volume overall and a high degree of
uncertainty as to when and at what level demand will occur, the time series is said to be lumpy,
or irregular”721. Among others many products and stocking locations might explain the lumpi-
ness of demand for individual Papierco products.722 “The lumpy demand condition occurs
when two or three times the standard deviation of the historical data exceed the forecast of the
best model that can be fit to the time series”723. Ballou (2003a: 9) states in the PowerPoint
presentation to Ballou (2004b) a time series is lumpy, if the mean of the series is less than or
equal to three times the standard deviation of the series.
Although the items in figure 5.3 are considered standard, on average they did not sell
8.29 weeks or around 16 % of the year 2007 and their standard deviation of sales per week
per warehouse was on average 0.94 times their mean. Fast moving copy papers selling
nearly every week were the least lumpy with the standard deviation being only 0.37 times
the mean on average. Uncoated Cut-Size White Wood-Pulp Paper was the lumpiest with
no sale in 34 weeks or almost two thirds of the year 2007 and a lumpiness factor (standard
deviation of demand divided by mean demand) of 2.11, followed by the Folding Box
Board with no sale in 15 weeks and a lumpiness factor of 1.32.
On average the standard deviation of an item's sale per month per warehouse is 2.24
times the mean sale per month per warehouse in 2007. Nearly one third (31.45 %) of the
products show a standard deviation that is more than three times the arithmetic mean of
sales per month per warehouse. In these cases the standard deviation is on average
3.78 times the mean. Of a lot of articles no unit is sold for several months and then sudden-
ly some month a lot of units are sold. On average 7.83 months or nearly two thirds of the
year 2007 a specific product is not sold. We conclude that demand per product per month
is very random and lumpy or sporadic and therefore “difficult to predict accurately by ma-
thematical methods due to the wide variability in the time series”724. Risk pooling methods
might produce relief here.
Ballou (2004b: 311) suggests to use the reasons for the lumpiness in the forecast, sepa-
rate forecasting of lumpy and regular demand products, not react quickly to random de-
mand shifts without assignable cause and use slowly reacting forecast methods such as
exponential smoothing with a low smoothing constant or a regression model refit annually
721
Ballou (2004b: 288). For further definitions please refer to e. g. Ghobbar and Friend (2003: 2105), Kukre-
ja and Schmidt (2005: 2059), Bridgefield (2006), Gutierrez et al. (2008: 409, 412).
722
Cf. Ballou (2004b: 288, 310).
723
Ballou (2004b: 310).
724
Ballou (2004b: 310).
79
at the most, and maybe make up for forecast inaccuracy by a higher inventory level for low
demand items. Croston (1972) and Johnston and Boylan (1996) recommend an exponential
smoothing constant of 0.05 to 0.20 for lumpy demand. Silver et al. (1998: 127), however,
report pertaining to intermittent and erratic demand that “[t]he exponential smoothing fore-
cast methods […] have been found to be ineffective where transactions (not necessarily
unit-sized) occur on a somewhat infrequent basis. […] Infrequent in the sense that the av-
erage time between transactions is considerably larger than the unit period, the latter being
the interval of forecast updating”.
Lumpy or highly uncertain demand can be coped with by obtaining information directly
from customers, collaborating with other supply chain and channel members, collaborative
forecasting, forecasting total demand and apportioning it by regions, delaying supply re-
sponse, product differentiation, and forecasting as long as possible, supplying the more
certain products first, developing quick response and flexible supply systems, and applying
the forecasting methods with caution.725 Most of these suggested solutions rely on risk
pooling.
5.2.4 Distribution Requirements Planning (DRP)

7000
6000
5000
4000 Inventory
Sales to Customers
Tons
Sales to Colleagues
Total Warehouse Sales
3000
Goods Received from Suppliers
Goods Received from Colleagues
Total Goods Received
2000
1000
0
Months

Warehouse Hemmingen
725
Ballou (2003a: 39-42, 2004b: 310f., 314-317).
80
6000
5000
4000
Inventory
Sales to Customers
Tons
3000 Sales to Colleagues
Total Warehouse Sales
Goods Received from Suppliers
Goods Received from Colleagues
2000
Total Goods Received
1000
0
Months

Warehouse Reinbek
Figures 5.4 and 5.5 illustrate Papierco's demand forecasting and inventory management prob-
lems using two standard Papierco warehouses. The inventory level is always around 1,000 to
2,000 tons higher than total warehouse sales. This may be partly explained by the large product
variety. The Total Goods Received and Total Warehouse Sales curves often are opposite in
phase or time-lagged, i. e. when total warehouse sales go up, often goods receipts go down or
vice versa.
Sales departments, product managers, and purchasing departments of the different
warehouse locations do not collaborate in planning demand and requirements. The sales
departments merely plan the customers' financial development, i. e. sales and contribution
margins per customer. On the basis of the product managers' aggregated sales planning on
the product group level for all German locations skeleton agreements are centrally nego-
tiated with the most important suppliers. The purchasing departments forecast demand on
the product level on the basis of the past six months' sales and order accordingly. These
three planning processes are unconnected. The sales departments do not report any ex-
pected exceptional demands to the purchasing departments either. The latter only plan
products sold via warehouses, without considering substitution effects between similar
81
products or between warehouse and drop-shipping sales726. Drop-shipping sales are con-
ducted only by the sales departments.
Papierco should create a uniform cross-functional demand planning process, in which
sales, product managers, and purchasing collaborate, to contain extraordinary demands
through the sales assistants' and product managers' profound market knowledge and enable
a better demand forecast and thus order policy. Although Adam and Ebert (1976) show
that objective statistical forecasting models produce more accurate results than subjective
judgemental forecasts, Bunn and Wright (1991) remark expert forecasts had a higher quali-
ty than previous research claims. That is why Silver et al. (1998: 133) suggest combining
these two approaches.
The individual locations conduct their distribution requirements planning (DRP) with-
out any coordination, so that inventory reductions at one location strain the transshipment
flows and impair other locations' DRP. Thus, one location might deplete another one's
stock of a specific product via transshipments. In the following week the latter location
might have to obtain this product via transshipments from another warehouse due to a cus-
tomer demand and so on. Likewise a warehouse may receive a supplier's delivery, although
it has plenty of inventory left, while another one does not. This process seems to have tak-
en on a life of its own in a “vicious circle” with increasing transshipment volumes and in-
ventory variability at most warehouse locations.
The increased transshipments did not reduce the aggregate safety stock, although this
might be assumed because of the demand balancing effect thanks to the creation of a vir-
tual inventory through transshipments. Allegedly the customer service level increased,
which is difficult to verify, as Papierco does not record it in any way.
We will recommend a not necessarily geographically centralized procurement, which
forecasts and orders for all of Papierco's German locations on the product level and directs
the suppliers' deliveries to the warehouses according to current demand in section 5.3.3.
5.2.5 Demand Forecasting Methods
Increasing forecast accuracy decreases the need for risk pooling727 (e. g. emergency transship-
ments) and safety stock and increases customer service level (product availability).
726
Sometimes a drop-shipping sale is converted into a warehouse sale, if the drop-shipping lead times are
high.
727
Cf. Graman and Magazine (2006: 1081).
82
Today Papierco uses the forecast methods 4 (Moving Average) and 10 (Linear Smooth-
ing) of the twelve ones the ERP software JD Edwards OneWorld offers.
Method 1 (Percent Over Last Year), method 2 (Calculated Percent Over Last Year), me-
thod 3 (Last Year To This Year) and method 8 (Flexible Method or Percent Over Months
Prior) are so called “naïve estimates” and project sales of previous years into the future
either directly or after multiplying them with a factor.728 These methods are rather inade-
quate for Papierco, as sales of individual products differ significantly from year to year.
Method 12 (Exponential Smoothing with Trends and Seasonality) is not suitable either, as
demand for individual products does not follow a trend or seasonal pattern.
Method 4 (Moving Average), Method 7 (Second Degree Approximation), which ex-
presses sales in a curve, method 9 (Weighted Moving Average), method 10 (Linear
Smoothing), which assigns weights to the sales numbers linearly decreasingly with a for-
mula, and method 11 (Exponential Smoothing729) are similar. They are suitable for short-
term forecasts for mature products without trends or seasonal patterns.730 They could ena-
ble a good demand forecast for Papierco with an order cycle of two weeks.
Method 5 (Linear Approximation) establishes a trend on the basis of two data points.
Small changes influence long-term forecasts strongly. Method 6 (Least Squares Regres-
sion) also calculates a linear trend. Turning points and changes in the demand function are
recognized late with the Linear Regression. If the sales data form a curve or show strong
seasonal fluctuations, systematic forecast errors appear.731 Methods 5 and 6 may be suita-
ble for Papierco because of the short order cycle of two weeks, although demand for indi-
vidual products shows no or only a short-term trend.
Analyzing the forecast methods for the seven standard products, we discovered that a
forecast based on sales of the last twelve months leads to improved results compared to the
current one with six months. This comes as no surprise as Gibson and Horner (2005: 1f.)
state that JD Edwards OneWorld forecasting accuracy improves with the length of the used
sales history. The software is capable of using up to two years of sales history.
Method 11 (Exponential Smoothing) and method 6 (Least Squares Regression) led to
the best results most frequently. Ballou (2004b: 311) also recommends these methods for
forecasting lumpy demand.
728
Oracle (2009).
729
Exponential Smoothing automatically uses linearly decreasing weights and is a weighted average of ac-
tual sales and the previous period's forecast.
730
Oracle (2009).
731
Oracle (2009).
83
The results of method 11 (Exponential Smoothing) might be improved by letting JD

Edwards OneWorld determine the smoothing constant alpha based on past sales instead of
determining it arbitrarily.
Methods 9 (Weighted Moving Average), 4 (Moving Average), and 10 (Linear Smooth-
ing) also led to good results in order of descending quality. A naïve estimation (projection
of the previous year's sales into the current year) did not give good results. The results
could be improved by multiplying the previous year's sales with a growth factor.
We used the mean absolute deviation (MAD) of the forecast values from the actual
sales in 2006 and 2007 and the forecast error (the sum of the difference of the actual and
the forecast sale per week) to measure the quality of the respective forecast method. For
other forecast performance measures please refer to Gutierrez et al. (2008: 413f.).
Comparison of the forecast accuracy of forecast methods is difficult732, as the forecast
errors of the two methods may be correlated733. Besides “a linear combination of forecasts
from two or more [forecasting] procedures can outperform all of those individual proce-
dures”734. According to Kang (1986) a regular average of forecasts of several methods
leads to high-quality results.
Tiacci and Saetta (2009) evaluate the impact of usually neglected possible interactions
between demand forecasting methods and stock control systems on global system perfor-
mance. Accordingly, demand forecasting methods cannot be chosen only on the basis of
traditional measures of forecast errors. Total costs and service level of the global inventory
control system should also be considered. However, this is beyond the scope of this thesis
and remains for further research.
The products still have a lumpy demand pattern also according to Ballou's (2004b: 310)
definition: “[T]he forecast of the best model that” could “be fit to the time series” was 3.39
times “the standard deviation of the historical data”.
The MAD was on average half the size of the mean actual demand per week, i. e. even
with the best tested forecast method, we would have ordered half a week's average demand
too much or too little per product per week on average. The absolute forecast error was on
average four times the mean actual demand per week, i. e. even though the forecast errors
balance each other to a certain extent we would have had four week's average demand too
much or too little over the whole year per product on average.
732
Silver et al. (1998: 131).
733
Peterson (1969).
734
Silver et al. (1998: 131).
84
Methods 11 (Exponential Smoothing), 6 (Least Squares Regression), and 9 (Weighted

Moving Average) should be admitted to the DRP forecast run besides today's exclusively
used methods 4 (Moving Average) and 10 (Linear Smoothing), as they gave good results.
Gibson and Horner (2005: 4) recommend using as many forecast methods as possible “to
maximize the chances that the Best Fit forecast that is calculated is really close to what
reality will be”. Of course, this has to be traded off with calculation time and software per-
formance.
“The statistical forecast techniques discussed earlier may work well for stable, mature
products with significant demand history, whereas qualitative methods based largely on
human opinion and direct marketing intelligence (not a part of JD Edwards) usually make
more sense for new or unstable demand products. […] In fact, conventional wisdom is that
sales and marketing should ʻown the forecastʼ. Their input is especially important for new
products or products with unstable demand. Sales and marketing should participate in the
development of forecasts and they should also be formally ʻgradedʼ on their forecast accu-
racy performance results. […] And one final comment: always remember that there is far
more to be gained by people collaborating and communicating well than there is by using
all of the advanced formulas and algorithms yet developed”735.
Since our analysis is only based on seven standard products, as no further data were
provided, it should be extended to a larger, more representative number of products before
implementation.
Simple forecasting methods outperform statistically sophisticated procedures for indi-
vidual items. The reverse is true for macro-level data and medium and long-term forecast-
ing. Consequently, sophisticated forecasting methods do not outperform simpler ones in
terms of forecast accuracy in general.736 Nevertheless, one could check whether more so-
phisticated forecasting methods not available in JD Edwards OneWorld and developed for
products with lumpy demand lead to more accurate results.
The Croston (1972) method forecasts intermittent demand well737 and is available in fore-
casting software packages738. It “generates separate forecasts for the demand size and the
number of periods between demands, and uses the ratio as an estimate for the expected de-
mand per period. It also updates the mean absolute deviation of the forecast error for the de-
735
Gibson and Horner (2005: 9).
736
Makridakis et al. (1982).
737
Willemain et al. (1994), Johnston and Boylan (1996).
738
Syntetos et al. (2005).
85
mand size”739. Teunter and Sani (2009: 82) call Croston's method “the standard method for
forecasting intermittent demand” and use its forecasts to calculate order-up-to levels that lead
to close to target customer service levels. Gutierrez et al. (2008) apply neural network mod-
eling to forecasting lumpy demand. Neural network forecasts generally outperform single
exponential smoothing, Croston's method, and the Syntetos–Boylan approximation740. Synte-
tos and Boylan (2001, 2005) approximately correct an error in the mathematical derivation of
and the resulting bias in Croston's estimates of expected demand.
Gutierrez et al. (2008: 418) find neural network models generally perform better than
the traditional methods. The Syntetos–Boylan approximation performs better than Cros-
ton's and single exponential smoothing methods in lumpy demand forecasting. If the aver-
age nonzero demand is considerably smaller in the test than in the training sample, the tra-
ditional forecast methods perform better than the neural network forecasts with lower
smoothing constants. However, Gutierrez et al. (2008: 412) only consider 24 SKUs.
Analyzing other methods for lumpy demand forecasting not available in JD Edwards
OneWorld lies beyond the scope of this thesis. Furthermore, it would be difficult and ex-
pensive to integrate these forecasting methods into the current ERP system and DRP.
A better demand forecast leads to a higher product availability (customer service level)
at the primary warehouse and thus reduces transshipments. Customer service level can be
increased and at least safety stock can be reduced.
Of course it would be desirable to always have the desired product at the right time at
the right warehouse, so that no transshipments would be necessary at all. This however is
impossible741, as forecasts are generally incorrect742. The longer the forecast horizon, the
worse the forecast is743. Aggregate are more precise than individual forecasts.744
As the suggested forecast methods in JD Edwards OneWorld only lead to a limited
forecast improvement and the more sophisticated methods for lumpy demand are difficult
to implement at least in the short run and their benefit for Papierco is unknown, this com-
pany could resort to risk pooling methods that reduce demand uncertainty relying on ag-
gregate forecasting745, such as central ordering.
739
Teunter and Sani (2009: 82).
740
Syntetos and Boylan (2001, 2005, 2006), Syntetos et al. (2005).
741
Evers (2001: 316).
742
743
744
Cf. footnote 456.
745
Sheffi (2004: 95).
86
5.3 Solving Papierco's Problems
Papierco could cope with lead time and demand uncertainty and its related forecasting and
inventory problems and ensure its competitiveness and economic success by risk pooling.
5.3.1 Determining Suitable Risk Pooling Methods for Papierco
The RPDST developed in section 4.4 is used to determine appropriate risk pooling methods for
Papierco, which suffers from customer demand and supplier lead time uncertainty (1). Both
demand and lead times fluctuate strongly.
Product variety is important (2) for Papierco's competitive advantage and because its
customers, who are mostly printing offices, need all kinds of paper specifications.
Stockout penalty costs are high compared to inventory carrying costs and thus attainable
inventory cost savings (3) from e. g. inventory pooling: The stockout cost (209 €) is esti-
mated by the average contribution margin of one ton of warehouse sales to customers mi-
nus the sales assistant's premium. The average length of time a ton stays in inventory is
estimated by the average inventory level divided by average annual demand. Papierco has
an annual average turnover rate of 8, i. e. on average a ton of paper remains in storage for
one eighth of a year or one and a half months. Storing one ton of paper costs 22 € per
month, so that the cost of holding a ton of paper in storage is 33 € on average.
The supplier lead time means and variabilities are high and therefore their reduction is
important to Papierco (4). Papierco declared costs per order did not accrue. However, if all
2007 expenses of the purchasing departments are divided by the total amount of warehouse
purchases from suppliers, we arrive at an order cost of 3.31 € per ordered ton. This cost per
order is relatively low (5) compared to the average value per ton of 1,058 € in 2007. De-
mand uncertainty and lead time (uncertainty) reduction and fewer orders are preferred to
only lead time pooling, but lower premium transportation cost and in-transit inventory cost
(15), as Papierco suffers from both demand and lead time uncertainty, minimum order and
saltus quantities have to be observed, and the in-transit carrying cost per unit and emergen-
cy transfer cost per unit are low.
Order splitting (16) is not feasible because of among others the required minimum order
quantities and full truck load supplier deliveries. The cheaper, faster implementable and
changeable, and less cross-functional solution to cope with both demand and lead time un-
certainty is preferred to product and/or process redesign of a postponement (14) strategy
(10). Besides, manufacturing and assembly postponement can rather be applied by the paper
producer, not the paper wholesaler. Papierco could postpone the labeling, packaging (ream
87
wrapping paper), or delivery of products, e. g. by central ordering the allocation of ordered

products can be delayed and conducted according to more current demand information. Mix-
ing printer's ink from base colors can also be postponed to the delivery warehouses until a
customer order arrives instead of ordering already mixed colors from the supplier. Cutting
paper to the desired format could also be delayed until an order arrives. However, this can
only be applied to an insignificant part of the product line. The other postponement options
for Papierco will not reduce demand and lead time uncertainty much either.
The paper producers do not possess small-order drop-shipping capabilities (11). They
only drop-ship full truck loads to customers. However, there are endeavors of some paper
wholesalers to conduct drop-shipping sales via its warehouses to gain the higher profit
margin of warehouse sales and of paper producers to drop-ship less than truck load to pa-
per wholesale customers to increase their profit. This causes resentment on both sides.
Virtual pooling (12) by not holding any inventory, only arranging sales, and having all
products drop-shipped from the paper producers to the customers is no viable strategy for
the paper wholesaler, as it would make him dispensable and the paper producers do not
possess the necessary distribution infrastructure. Furthermore, the restriction to deliver to
customers usually within 24 hours could not be kept. Therefore lateral transshipments
(13) or cross-filling seem to be the risk pooling method of choice for Papierco to deal with
both demand and lead time uncertainty.
Order costs may also be considered high in (5), because minimum order and saltus
quantities make (placing) an order more expensive, as frequently more than forecast de-
mand minus on hand inventory has to be ordered by the individual locations to meet these
order restrictions, which results in unnecessary inventory costs. Increasing the needed or-
der quantity per order by postponement (8) or central ordering (9) for several locations
may better exhaust minimum order and saltus quantities and lead to economies of scale in
ordering (6) due to price discounts. As Papierco prefers the less expensive, fast implement-
able and changeable solution and the possibilities for postponement (8) are still limited as
described above, central ordering (9) might be a viable risk pooling method to achieve both
efficiency and responsiveness to customers. Some of the transportation cost otherwise in-
curred for emergency transshipments between warehouses may be shifted unto the suppli-
ers who deliver to Papierco's delivery warehouses. Therefore (9) central ordering might
make sense, especially against the background of increasing fuel prices.
Today Papierco's average transportation cost to deliver one ton of paper with its truck
fleet within Germany is 15.88 €. We arrived at this figure by dividing Papierco Germany's
88
total transportation cost (driver costs, capital commitment, depreciation, maintenance and
repair, motor vehicle tax, insurance, diesel, lubricant, tires, autobahn toll, parking, and
miscellaneous operating costs) by the total annual warehouse sales in 2008. The trans-
shipment cost per ton (transportation, shipment documentation, receiving, handling and
administration costs) amount to 69.48 € per transshipped ton of paper on average. This is
still relatively moderate (6), so that the decision process (10) through (13) can be followed
again, which reconfirms transshipments (13) as a favorable risk pooling method.
Papierco's management thinks that (2) product variety might be reduced in the office
paper segment and where product specifications just differ in their format in one or two
centimeters. The Paper Technology Research Association Forschungsvereinigung Papier-
technik e. V. (FPT) also denies the necessity of the current large product variety in German
paper wholesale.746 Affirming the importance of lead time (uncertainty) reduction (30)
afterwards leads to the same train of thoughts in (31) to (38) as in (5) to (16) without con-
sidering postponement (cf. 2), which favors transshipments (36) and perhaps central or-
dering (33).
If Papierco gives priority to reducing demand uncertainty, as demand variability is more
severe and lead times and their variability are not recorded accurately nowadays, and there-
fore (30) is denied and accommodating demand uncertainty is preferred to inventory reduc-
tion (39) due to the fierce competition for customers, then central ordering (33) is favored
again: It is less expensive, faster implementable and changeable than capacity pooling
(41f.) and permits to achieve economies of scale in ordering (40). Besides, capacity pool-
ing is rather suitable for manufacturing companies than for wholesalers. Papierco's loca-
tions cannot pool manufacturing capacity, as they do not produce paper. They could, how-
ever, pool or share printer's ink mixing, paper cutting, or ream wrapping capacity. This
might not be economically worthwhile, since the costs of shipping mixed paint or cut to fit
or ream wrapped paper between locations might destroy the capacity pooling benefits. Fur-
thermore, delivery time to the customer might suffer, and this can only be applied to an
insignificant part of the product line. Papierco's locations could pool transportation capaci-
ty, which is operationalized by transshipments, and inventory storage capacity, which cor-
responds to inventory pooling.
Papierco's management would also like to reduce inventory. One has to carefully check
whether this intention does not conflict with Papierco's business strategy and competitive
746
FTP (2007a, 2007b).
89
advantage of being a 24-hour delivery full-range paper wholesaler in Germany and inven-
tory savings outweigh possible lost sales. Remember that Papierco's inventory carrying
costs are moderate relative to stockout and transshipment costs per ton. If one concludes
that inventory reduction is more important than accommodating demand uncertainty (39),
product pooling (45) and inventory pooling (46) might be suitable. Product pooling
keeps inventory close to customers in the delivery warehouses, thus enables the nationwide
24-hour delivery service and product availability (44), pools demands, and improves fore-
cast accuracy, but perhaps degrades product functionality. Because of the mentioned short-
comings and because it only pools demands, product pooling should not be applied on its
own but together with another risk pooling method that reduces lead time uncertainty and
only within a product segment that can be rationalized without hurting customer service
level and product purchase interdependencies substantially. Rationalization in this context
means deleting products from the product line or replacing several product variants by a
universal product.
Physical inventory pooling in a single central warehouse would preclude the 24-hour de-
livery service. However, selective stock-keeping in several delivery warehouses may be feas-
ible, pool demands, enable to reduce at least safety stock and maintain the product variety
and delivery speed, but increase transportation costs. Fast movers could be stored at every
warehouse, while slow movers could be held at a limited number of warehouses only.
If Papierco disavows the importance of lead time (uncertainty) reduction in (30), it may
also do so in (4). Papierco's products share common qualities (47). The less expensive, fast
implementable and changeable solution is preferred to product and/or process redesign and
leagilty, and Papierco is a trading company (48). It does not produce its paper products
itself and its products neither are modular nor do they share common components besides
the raw materials, so that it cannot redesign the products and/or production processes.
Hence, component commonality (54) or manufacturing postponement (53) is not sensible
for Papierco. Postponing cutting paper to the right format, mixing printer's ink, and labe-
ling and ream wrapping paper are conceivable but not expected to reduce demand or lead
time uncertainty significantly.
Product substitution (50) can be implemented quickly and cheaply, but displease cus-
tomers as they do not receive the desired product in case of a stockout, but a substitute. In
contrast to product substitution, central ordering (33) might reduce lead time (uncertain-
ty) in addition to demand uncertainty and lead to economies of scale in ordering, but be
more expensive, particularly since transaction and coordination costs are higher.
90
On the whole, lateral emergency transshipments seem to be the most suitable risk pool-
ing method for Papierco, followed by central ordering, especially if fuel and transportation
costs continue to rise. Both methods may be applied in combination and enable to achieve
the seemingly conflicting aims of reducing demand and lead time uncertainty as well as
inventory investment and maintaining customer service level in terms of product variety,
inventory availability, and delivery speed. Product substitution, product pooling, inventory
pooling, and postponement may also be suitable in order of descending favorability, but
not as a sole method due to their mentioned drawbacks. They may not reduce lead times or
lead time uncertainty by themselves.
In the following we will deal with the suitable risk pooling methods determined for Pa-
pierco in more detail.
5.3.2 Emergency Transshipments between Papierco's German Locations
Papierco's German locations exchange products in case of a stockout, which supports the result
of the RPDST.
As the quantities exchanged by transshipments have risen over the last years, Papierco's
management asked whether these transshipments are economically sound and whether they
can be reduced or improved.
In our opinion this stock exchange must not be considered detachedly from sales747,
demand, demand forecast, order policy and storage policy, as it is influenced by these fac-
tors. Therefore the systemic approach of logistics should be accommodated and the inter-
dependencies between single components and the whole should be considered.748 It is de-
sired to contribute to a preferably spatially and temporally smooth, continuous, and aligned
sequence of activities and processes which target satisfying customer needs.749
5.3.2.1 Optimizing Catchment Areas

A business policy has evolved historically that if the sales department of one location attends to
a customer, the same location's logistics department also delivers products to this customer.
Therefore, today some of Papierco's customers in the same five-digit zip code areas are allo-
cated inefficiently to several of the current 21 German warehouse locations. Customers should,
however, be supplied standardly by the warehouse closest to them. There is a high potential for
747
The sales assistant triggers a transshipment order, i. e. he orders stock from another warehouse, if he
cannot satisfy a customer's demand from the primary warehouse.
748
Delfmann (2000: 322).
749
Delfmann (2000: 325).
91
transportation cost degression in switching from today's suboptimal to the determined optimal
customer allocation to Papierco's current 21 warehouses. Unfortunately, it cannot be numera-
lized exactly due to today's inefficient customer allocation to multiple locations, erroneous data
management, and ignorance of the number of delivery trips to each customer, delivery routes
and quantities, and truck utilization before and after optimization.
Optimizing the customer allocations can lead to cost reductions fast and without much
investment burden. The delivery route numbers merely have to be changed for some cus-
tomers in the customer master data of the ERP software. Furthermore, optimal customer
allocations may reduce transshipments.
The optimal customer allocation according to the shortest distance was found with the
software NCloc750 and Euclidean metric and visualized with the zip code diagram Das
Postleitzahlen-Diagramm 3.7751 in figure 5.6. The colored areas represent today's, the ones
outlined in black the optimal catchment areas.
750
ATL (2009).
751
Wessiepe (2009).
92
Figure 5.6: Comparison of Current and Optimal Catchment Areas of Papierco's Warehouses
93
5.3.2.2 Increase in Transshipments and Its Causes
Year Warehouse Sales Transshipments (t) Transshipments

to Customers (t) (% of Sales to Customers)
2005 381,694 53,636 14.05
2006 405,116 58,755 14.50
2007 428,491 71,967 16.80
Table 5.1: The Extent of Transshipments at Papierco
We can confirm that the transshipment quantities rose from 2005 to 2007 by 34.18 % (from
2005 to 2006 by 9.54 % and from 2006 to 2007 by 22.49 %). The warehouse sales to custom-
ers also increased from 2005 to 2007 by 12.26 % (from 2005 to 2006 by 6.14 % and from
2006 to 2007 by 5.77 %). However, this does not fully explain the increase in transshipments,
as the percentage of warehouse sales to customers enabled by transshipments increased as well
(cf. table 5.1).
In interviews with chief operating officers and directors of purchasing, sales, and logis-
tics at Papierco's different locations it became clear that the increase in transshipments is
caused by changing transshipment operations to direct invoicing in June of 2006752, some
locations decreasing their inventory levels, and including more and more products in the
product line753. According to one chief executive officer the “dramatic increase” in trans-
shipment volumes was “the manifestation of the sales department's service delusion”.
However, thanks to transshipments Papierco had also gained market share in warehouse
sales. Papierco's better development compared to its competitors was based on transship-
ments, as they enabled a higher product availability, broader product assortment, and faster
delivery. This confirms that transshipments can lead to a competitive advantage as Gug-
lielmo (1999), Kroll (2006), and Alvarez (2007) already insinuate.
The cost that locations are charged for using an emergency transshipment (transship-
ment cost rate) is based on quantity units (positions, pallets, and kilograms) and purchase
prices and does not reflect the actual logistical costs (picking, transportation, and reloading
costs). Within a company group only the costs really incurred by the transshipments should
752
Sales people do not search at the own warehouse for an alternative product to the one originally desired
by the customer in case of a stockout anymore. Today sales assistants order more via transshipments for
reasons of simplicity than prior to the introduction of direct invoicing. This puts a strain on the product
exchange flows between warehouses in terms of volume and costs.
753
These products tend to have small sales volumes and to steal sales from substitutes. Nonetheless, at least
small quantities have to be stored at every location, which increases average stock, or the product has to
be obtained via transshipments, which increases transshipment quantities.
94
be charged, so that there is no incentive to prefer or reject the use of transshipments due to
inadequate charges.
Although Rudi et al. (2001) find transshipment prices that are supposed to always in-
duce local deciders to make globally optimal inventory and transshipment decisions, Hu et
al. (2007) give a counterexample. There may exist coordinating prices for only a small
range of problem parameters, e. g. if the two locations are symmetric (both locations have
the same cost and revenue coefficients, demand distributions, and capacity constraints, and
changes in either demand or capacity distributions are simultaneously implemented in both
locations). The higher the capacity uncertainty is, the higher the coordinating prices are.
Increasing demand variability may increase or decrease coordinating prices depending on
the distribution. A linear transshipment price schedule often will not achieve coordination
of two locations' production and transshipment.
Apart from this the transshipment lead time tends to be shorter than the supplier lead
time. In Papierco's case the transshipment lead time ranges from a few hours to 4 work
days at the most. The supplier lead time for the various products ranges from 14 to 154
days and is 28.49 days on average. Therefore a transshipment from a sister location is pre-
ferred to a replenishment from a supplier in case of a stockout, as otherwise the sale is lost
due to the usual delivery time restriction of 24 hours.
Some managers remark that for some Papierco locations other ones were their “best
customers“, as they made profits by transshipping products. One director of logistics
claimed only locations obtaining a lot of products via transshipments criticized the trans-
shipment charge rate as too high. It would be even higher, if the really incurred logistical
costs were charged. This supports the call for fair transshipment charge rates.
The transshipment costs ought to be debited to the sales assistants (deducted from the
contribution margin which determines the sales assistants' premium) who release a trans-
shipment order or at least made transparent to them. The objection that one should not in-
hibit the sales assistants in their sales function can be countered that other wholesalers
commonly set up guidelines for the sales personnel.
Furthermore, if a sales assistant chooses, for instance, a paper with a one centimeter
larger format as an alternative to a stocked out product desired by a customer, the higher
purchase price cannot be charged. Therefore, the contribution margin that determines the
sales assistant's premium is lower and the sales assistant is enticed to obtain the originally
demanded product via transshipment from another warehouse location. This highlights that
the implementation of risk pooling methods is influenced by human resource and business
95
policy factors. Managers should formulate clear guidelines for the usage of transshipments
in case a desired item is not available at the primary stocking location.
A software tool should be created to help the sales assistant in deciding whether it is
more economical to obtain a stocked out item via transshipments from another warehouse,
search for a substitute at the own warehouse, cut paper to the right format, or ream wrap
paper at the own location if possible.
5.3.2.3 Transshipments Are Worthwhile for Papierco

A high customer service level at the warehouses can be achieved by either holding sufficient
inventory there, which entails high inventory investment and carrying costs, or by postpone-
ment through centralized inventory holding, which presumes fast and thus expensive delivery
of any amount of any product to any customer. A better alternative might be to hold some in-
ventory at the warehouses and let the warehouses transship products in case of a stockout.754
“A mathematical model of the exact implications [… of transshipments] is extremely
complicated”755. Evers (2001), Minner et al. (2003), and Minner and Silver (2005) consider
transshipments between two locations for a specific product in a specific situation only and
not whether transshipments might be worthwhile for a company with more locations and
products in general at all. Minner et al. (2003: 394) note that their model may be extended
to multiple locations with no fixed transshipment costs by successively determining which
location maximizes the savings for each transshipped unit. The problem becomes more
difficult with fixed costs, unless there are only a small number of warehouses as in Tagaras
(1999). The structures in all past transshipment research are much simpler than the practi-
cal ones.756
If the outlets use complete pooling, the exact form of the lateral transshipment policy
does not significantly affect total system performance.757 Deciding between never trans-
shipping or always transshipping as many units as disposable if one location faces a stock-
out while another one has stock on hand758 is almost as good as a more extensive investiga-
tion that considers the transshipment's prospective effect on the transshipping location's
cost759. Burton and Banerjee (2005: 169) also find that an ad hoc emergency transshipment
approach may lead to lower shortage (higher customer service) and transshipment cost
754
Cf. Evers (2001).
755
Minner et al. (2003: 385).
756
Chiou (2008: 442).
757
Tagaras (1999).
758
Cf. Olsson (2009).
759
Minner et al. (2003: 394f.), Minner and Silver (2005: 1).
96
(number of lateral transshipments) than a more systematic transshipment technique based

on stock level equalization760. Transshipping is better than not transshipping, but it increas-
es transportation activity.761 Backordered demand should not be partially fulfilled by trans-
shipments.762 In practice pull systems may offer simpler and more practical joint reple-
nishment/transshipment policies than the push systems mostly considered in the litera-
ture.763
At Papierco the charged transshipment costs (for transportation, shipment documenta-
tion, receiving, handling, and administration) amounted to 5 Mio. € in 2007, i. e. about
69.48 € per ton that was delivered via transshipments (5 Mio. €/71,967 t = 69.48 €/t). Ac-
cording to its inventory turnover curve (figure 5.7), Papierco needs around 10,291 t of av-
erage inventory for 71,967 t of warehouse sales to customers. This suggests that without
transshipments Papierco would have had to store at least 10,291 t of additional average
inventory.
This derivation might not be fully justified, as we derive inventory figures for the sys-
tem without transshipments from the figures for the existing system with transshipments
and while safety stocks will be lower in general, regular stocks may increase with cross-
filling of demand764.
Before transshipments were used, average inventory turnover in German paper whole-
sale was 5.4 from 1975 to 1984765. Therefore one might estimate that Papierco would have
needed 13,327 t (= 71,967 t/5.4) extra average inventory without transshipments. Conse-
quently, Papierco would have incurred approximately 3.5 Mio. € of additional inventory
carrying costs in order to sell 71,967 t without transshipments: 13,327 t × 1,058 €/t (av-
erage product purchase price in the warehouse sales business division) × 0.25 (average
inventory carrying cost rate) = 3,524,992 €. However, these additionally stored tons would
not necessarily have been the demanded ones in the right location. As the estimate of addi-
tional inventory carrying costs without transshipments is lower than the 2007 total trans-
shipment cost, the latter should be reduced, e. g. by storing fast-movers such as copy pa-
pers at all locations (cf. section 5.3.5), avoiding unnecessary and uneconomical transship-
ments, improving demand forecasts, and using less expensive alternative risk pooling me-
thods.
760
See, e.g., Jönsson and Silver (1987a).
761
Burton and Banerjee (2005: 169).
762
Wee and Dada (2005: 1529).
763
Bhaumik and Kataria (2006).
764
765
Röttgen (1987: 109).
97
Centralizing all inventories in one or several central warehouses would decrease ware-
house fixed costs, safety stock and thus inventory holding costs as well as inbound trans-
portation costs to these central warehouses, but also the customer service level due to
greater delivery distances to the customers (the stockout costs would be higher and the
profit possibly lower) and increase outbound transportation costs to customers. Besides, a
central warehousing strategy would conflict with Papierco's corporate strategy of a decen-
tralized full-line 24-hour-delivery distributor.
As stockout costs could not be estimated in the aggregate consideration without trans-
shipments in general, we now consider the costs of transshipping versus not transshipping.
Transshipments entail transportation, shipment documentation, receiving, handling and
administration costs, the sending location's cost of reordering the transshipped product
from the supplier, as well as an increased probability of a stockout at the sending loca-
tion.766
Transportation, shipment documentation, receiving, handling, and administration costs
amount to 69.48 € per transshipped ton of paper on average. As derived in section 5.3.1 the
order cost per ton amounts to 3.31 €. Even though Papierco's locations order and receive
stock regularly about every two weeks and thus it is unlikely to place an order merely to
make up for transshipped items, we consider this value as an upper bound to the order cost
caused by transshipping.
The increased probability of a stockout at the sending location is difficult to quantify in
general and on average. In Papierco's case this cost is rather another transshipping cost
again, as the sending location would ask another location for a transshipment, if later it ran
out of stock of the product it had just transshipped. The increased stockout likelihood de-
pends on the transshipment quantity, the current on-hand inventory of the transshipped
product, customer demand for this item, and the stock receipts of this item (quantity on
order and its arrival time).
Transshipments made up 16.8 % of warehouse sales to customers in 2007, i. e. one could
assume that Papierco had an average overall stockout probability of 16.8 % and therefore an
inventory availability or fill rate before transshipments of 83.2 %. We assume that every
location faces a stockout probability of 16.8 % for an item, neglecting the different influen-
cing factors just mentioned, the derivation of this stockout probability from the situation with
transshipments, and the difference between alpha and beta service level. If two locations
766
Evers (2001: 312f.).
98
transship, they face a stockout with a probability of 0.168 100 2.82 %, three with a
probability of 0.168 100 0.47 %, four with 0.08 %, five with 0.013 %, six with
0.002 %, seven with 0.0004 %, and 21 with 5.3888354 10 %. This affirms that emer-
gency transshipments lead to a high effective fill rate for the customer (97.18 % for two,
99.53 % for three, and 99.92 % for four locations that transship), even if the item fill rate is
relatively low (83.2 %).767 The main benefit (reduced stockouts and safety stock levels) can
be reaped having just a few locations transship. It is neither necessary nor economical to
have all 21 locations transship due to diminishing marginal returns to transshipping768.
If we consider the worst but unlikely case that at every location one after another a
stockout occurs for the same item after having transshipped one ton of that item to the pre-
ceding location, then the reorder cost due to transshipment only accrues at the last (21st)
location. The cost of increased probability of a stockout can be estimated by another trans-
shipping cost of 69.48 €, since the likelihood that one transshipment triggers more than one
additional stockout and thus transshipment is negligible as the example above shows.
Therefore the average cost of transshipping can be estimated as: 142.27 €/t = 69.48 €/t
(transportation and related costs) + 3.31 €/t (reorder cost) + 69.48 €/t (increased transship-
ment usage/stockout probability).
Without transshipments Papierco incurs the cost of a stockout and the cost of holding a
ton in inventory until later at the not transshipping location. The stockout cost (209 €) is
estimated by the average contribution margin of one ton of warehouse sales to customers
minus the sales premium. The impact of stockouts on the future purchasing behavior of
customers is difficult to grasp and neglected here. Holding the ton of paper in storage until
later costs 33 € on average (cf. section 5.3.1). Therefore, the cost of not transshipping
amounts to approximately 242 €/t.
Transshipping is on average economically sensible for Papierco, as it generally is less
expensive (142.27 €) than not transshipping (242 €) a ton of paper in case of a stockout.
If only transportation-related costs evoked by transshipments and stockout costs due to
not transshipping are considered and the other costs are neglected, a sales assistant should
only make a transshipment order in case of a stockout, if the contribution margin is equal
to or larger than 69.48 €/t. Then at least the average transportation-related costs for the
transshipment are covered.
767
Cf. Ballou and Burnetas (2000, 2003), Ballou (2003b: 81, 2004b: 385-389).
768
Tagaras (1989), Evers (1997: 71f.), Ballou and Burnetas (2000, 2003), Ballou (2004b: 385-389), Kranen-
burg and Van Houtum (2009).
99
In order to decide more accurately on emergency transshipments for a specific item in a

specific situation the methods proposed by Evers (2001), Minner et al. (2003), and Minner
and Silver (2005) can be used, provided that the necessary data are available. Minner et al.
(2003: 395) present a heuristic decision rule for the quantity and source of a transshipment.
In simulations, Minner et al.'s (2003: 390f., 393) method generally achieves lower average
annual costs than Evers' (2001) one. Since the cost parameters are likely to remain constant
in the short run, whether or not a particular transshipment is made depends principally on
the on-hand inventory level at the shipping location.769
However, before using transshipments, sales assistants should check their own ware-
house for substitutes for the customer wish and try to persuade the customer to buy another
product, which only incurs transaction costs (searching for substitutes in the information
system). This product substitution might cause a slight dissatisfaction on the customer's
part, but is certainly cheaper than a transshipment. If the alternative product is more expen-
sive than the original customer wish, a higher profit is gained. If only the sales price for the
originally desired product can be charged or the substitute is less expensive than the origi-
nal purchase wish, Papierco has to accept a lower contribution margin. Of course the prod-
uct should not be sold beneath cost price.
If no substitute can be found, it should be checked whether the paper can be cut to the
right size (form postponement770), which costs around 56.25 €/t (neglecting waste), or
ream wrapped (packaging postponement) for 33.50 €/t on average. Both options are
cheaper than and therefore should be preferred to a transshipment in general.
In order of increasing costs incurred by the considered risk pooling methods, Papierco's
sales assistants should first try to substitute an unavailable product, then to customize or
differentiate products, and then to resort to transshipments in case of a stockout. This
shows that risk pooling methods can be combined effectively.
Although this is a rather crude general average analysis, because exact data (especially
probabilities) were not available, and we advert to the perils of using averages as noted
e. g. by Savage et al. (2006), it shows some general relationships. Of course a decision for
a particular risk pooling method (product substitution, postponement, or transshipment) in
a certain situation depends on the specific cost parameters.
769
Evers (2001: 313).
770
See e. g. Chauhan et al. (2008).
100
5.3.3 Centralized Ordering
5.3.3.1 Papierco's Current Order Policy

At Papierco a replenishment order is placed with a supplier (the daily DRP run generates an
order message) whenever the reorder point is reached or in possibly short intervals (every two
weeks) within the suppliers' minimum order quantity restrictions. The reorder point is the sum
of safety stock, which is set by every Papierco member company independently, and average
lead time demand (average sales per day during the lead time times the lead time in days).
The purchase order planning code for a specific product is 1, if it is planned with DRP,
and 0, if it is not planned with DRP. The latter case applies to about 20 % of the warehouse
products, as for them the minimum order quantity is usually higher than the planned order
quantity, e. g. if they have fixed order dates or the minimum order quantity is a full truck
load. These products are planned manually with order lists.
Safety stock is calculated as the product of average sales per day during the lead time,
supplier lead time in days, and a safety factor. Thus double average demand during lead
time results in double safety stock. According to experience the safety factor is set to ¼ for
ream wrapped paper and 1/3 for not ream wrapped one. The safety factor is not entered
into the ERP system for every product, but for parts of the product line manually every
three months.
However, safety stock should not be set arbitrarily, but according to a target customer
service level (e. g. 90 % product availability for A items) and the statistical behavior of
demand during lead time or rather the forecast error in the case of Papierco, because de-
mand is forecast. The actual supplier lead time and whether and how often the safety stock
is touched is not recorded either. No cost optimal order quantities or times are determined.
Folding Box Board (cf. figure 5.3) has an average demand during one week lead time of
819.23 sheets, a current safety stock of 204.81 (= 819.23 × 1/4) sheets, and a reorder point of
1024.04 (= 819.23 + 204.81) sheets. According to the empirical distribution function this cor-
responds to a service level of approximately 76 % or a stockout probability during lead time of
24 %. A reorder point of 1.460 sheets or safety stock of 640.77 (= 1460 - 819.23) sheets would
be needed to reach an inventory availability of 80.8 % during lead time. According to the re-
spective empirical distribution functions, the Uncoated Cut-Size White Wood-Pulp Paper has an
inventory availability during lead time of 86 %, one continuous roll of approximately 84 %, the
other of 73.1 %, Woodfree Matte Coated Cut-Size Paper of 72 %, the Woodfree White Cut-
Size C-Quality Copy Paper of 76.9 %, and the Woodfree White Cut-Size A-Quality Copy Paper
101
of 86.5 % with today's calculation of safety stock. The safety stock can be determined more
definedly based on a target customer service level.
According to the nonparametric Kolmogorov–Smirnov test for the second continuous
roll (asymptotic significance (two-tailed) p = 0.700) and the copy papers (p = 0.741 and
0.616) the hypothesis H0 that the sample stems from a population with normally distributed
weekly demands cannot be rejected with a level of significance of 5 %. The Kolmogorov–
Smirnov test is also applicable to small samples, but might not be very definite if the un-
known distribution function is not continuous (as in this case) or the data are not metric.771
The hypothesis H0 that the sample stems from a population with uniformly, Poisson, or
exponentially distributed weekly demands is rejected with a level of significance of 5 %
for all products but the second continuous roll (for the exponential distribution p = 0.436).
However, there is one value outside the specified exponential distribution range that is
skipped. In the Kolmogorov–Smirnov test of exponential distribution for the other paper
grades (variables) there are also values outside the specified distribution range that are
skipped.
The used Order Policy Code 4 (Periods of supply) combines the order quantities (fore-
cast demands) for the time period defined in the Value Order Policy (for most Papierco
products 10 days). The JD Edwards OneWorld manual says “This order policy code was
designed to use with high-use/low cost items for which the user is not concerned about
carrying excess inventory”772. This partly explains why Papierco carries so much invento-
ry, although it mainly sells low-turnover and not necessarily low-cost products.
The other available Order Policy Codes are 0 (The DRP system will not plan this item.),
1 Lot-for-lot or as required (The system will create messages for purchase orders for the
exact amount needed.), 2 Fixed order quantity (The system will create messages for pur-
chase orders based on the value entered in the Value Order Policy in the Plant Manufactur-
ing Data. If demand exceeds the fixed order quantity, the system will generate purchase
orders in multiples of the fixed order quantity.), 3 Economic Order Quantity (EOQ), and 5
Rate scheduled item (The system will generate messages based on existing rates and rate
generation rules for parent rate-scheduled items only.).
The order quantity is the maximum of the ten day demand forecast and the minimum
order quantity, which is rounded to the next higher order quantity that is divisible by the
saltus quantity. For instance, if the minimum order and saltus quantity are 500 kg (only
771
Duller (2008: 108, 112).
772
Oracle (2009).
102
complete pallets must be ordered) and forecast demand for ten days is 1,150 kg, then
1,500 kg are ordered. Such a policy that determines and controls the inventory level at each
warehouse location in direct proportion to (forecast) demand is called stock-to-demand.773
Papierco's inventory turnover curve in grey in figure 5.7 confirms this. For all its
warehouses the average annual inventory in dependence of the annual throughput is plotted
with green squares. A linear regression line is adequate for these data as the coefficient of
determination R2 = 0.8362 shows. However, it also suggests that not all warehouses ex-
ecute this order policy consistently.774 Individual turnover rates range from 5 to 21 with an
average of 8.24. Deviations of the individual turnover rates might be explained by different
service levels or replenishment rules.775 Warehouses, whose data points lie above the turn-
over curve, might attain a higher turnover and lower inventory holding costs, if they fol-
lowed the stock-to-demand order policy more consistently. Warehouses, whose data points
are plotted beneath the turnover curve, exhibit a higher-than-average turnover rate than
attainable by this stock-to-demand policy.
An EOQ policy would lower the average turnover rate to 6.53 as the turnover curve in
black shows (a power function with the normative exponent suggested by Ballou (2000:
75) fitted to Papierco's data), probably because the stock-to-demand order policy is not
followed consistently today and some warehouses have an unusually high turnover rate of
11, 15, or 21.
Nevertheless, this paper distributor's average inventory levels appear rather high for the
annual throughputs. The annual average turnover rate of the German wholesale of paper,
cardboard, and office supplies from 1999 to 2006 was 17.3.776 However, this group con-
tains stationary wholesalers, which might raise the average turnover rate and make it un-
suitable as a benchmark for just paper wholesalers. The U.S. merchant wholesalers of pa-
per and paper products had an annual average turnover rate from 1992 to 2008 of 12.75.777
Papierco has a rather low average turnover rate of 8.24. This might be explained by the
large product variety it offers. Papierco sells at least 7,000 different product specifications.
Thus it is difficult to forecast demand for individual products. Furthermore, minimum or-
der and saltus quantities have to be met.
773
Ballou (2000: 74).
774
Cf. Ballou (2000: 76).
775
Cf. Ballou (2000: 78f.).
776
Statistisches Bundesamt Deutschland (2009a, 2009b, 2009c, 2009d, 2009e, 2009f, 2009g, 2009h), calcu-
lated by us.
777
U.S. Census Bureau (2009), calculated by us.
103
10000
9000
y = 0.143x
8000 R² = 0.8362
Average Inventory (t)
7000
y = 3.1756x 0,7
6000
5000
4000
3000
2000
1000
0
0 10000 20000 30000 40000 50000 60000 70000
Annual Warehouse Throughput (t)
EOQ Current Stock-to-Demand Policy Pot.(EOQ) Linear (Current Stock-to-Demand Policy)
Figure 5.7: Inventory Turnover Curves for Papierco

EOQ = economic order quantity, Pot. = regression power function, Linear = linear regression line.
Papierco has 196 sum suppliers, but 83 % of the annual procurement quantity is sourced from
ten, 52 % from three, 40 % from two suppliers, and 20 % just from a single supplier. With big
suppliers a quantity structure is negotiated. There is a one-to-one product-supplier relationship,
i. e. every product is sourced from only exactly one supplier.
Therefore, Papierco should collaborate closer with its suppliers, conduct a better sup-
plier management, to enable shorter lead times, a better forecast, inventory policy and
product availability. One Papierco member company's director of purchasing had created
such a collaboration with Papierco's most important supplier that supplied the highest
quantity. A monthly demand forecasting for product groups and a guaranteed maximum
lead time of 14 days for fast and 28 days for slow movers was established. Previously this
supplier's lead time had been 4 to 6, sometimes 8 to 12 weeks. Inventories could be re-
duced considerably. Unfortunately, the business relationship to this supplier was broken
due to price policy reasons. This concept was neglected, and inventories rose again.
One director of purchasing is against a partner-like collaboration with suppliers for they
already took advantage of Papierco due to their big negotiation power. One director of
sales remarked, paper wholesalers and manufacturers did not cope with recurring demand
peaks correctly, such as in election times. When there were vacation close-downs, they
could not prevent supply bottlenecks. This shows again that a close collaboration with
104
suppliers is indispensable. Not least Weng and McClurg (2003) showed the value of sup-
plier–buyer coordination (in ordering) in coping with uncertain delivery time and demand.
5.3.3.2 Stock-to-Demand Order Policy with Centralized Ordering and Minimum

Order and Saltus Quantities
Risk pooling models are based on economic-order-quantity (trading off inventory holding
against order costs) or newsvendor models (trading off overage against underage costs).
In practice, however, a stock-to-demand order policy is followed frequently778 as we al-
so detected in interviews with logistics and purchasing managers and at Papierco. Orders
and therefore cycle inventories are proportional to demand at each warehouse.779 Expected
demand at each warehouse location is forecast. Order quantities result from the difference
of forecast demand and available inventory. Maister (1976: 132) and Evers (1995: 5) refer
to this order policy as the “replacement principle”, where orders are set equal to demand
during a certain time period. Such an order policy leads to a constant turnover. It is fre-
quently pursued, as it is simple and easily executable780 and because automated ordering
systems have very low ordering costs and thus the EOQ is not used781. The replacement
principle is followed e. g. in Zinn et al. (1990: 139), Evers (1996: 117, 1997: 59), and Ra-
binovich and Evers (2003a: 226).
If it is applied, centralization usually does not reduce cycle stocks:782 The system wide
order quantity of n warehouses and therefore average cycle stock in the decentralized sys-
tem are equal to the ones in the centralized system, if exactly the forecast demand is or-
dered.
Following a stock-to-demand order policy, safety stock is often calculated as the prod-
uct of a safety factor and average demand during lead time at each warehouse as our inter-
views with logistics and purchasing managers showed. Doubling demand results in double
safety stock. Safety stock is not determined by a desired target customer service level and
the statistical behavior of demand during lead time or the forecast error. Centralization
does not reduce system wide safety stock, if the safety factor is equal for all warehouses
778
Wanke (2009: 108).
779
Ballou (2000: 74).
780
Ballou (2000: 77f., 2004a: 20f.).
781
Herron (1987), Evers (1995: 14). Researchers disagree whether the EOQ model is rarely used (Herron
1987, Woolsey 1988, Zinn et al. 1990: 139, Evers 1995: 14) or often used (Osteryoung et al. 1986, Pan
and Liao 1989, Lee and Nahmias 1993: 48, Dubelaar et al. 2001: 98, Slack et al. 2004). Of eleven com-
panies that we surveyed (table D.1) five use a stock-to-demand, four an EOQ, one a dynamic order poli-
cy, and one the newsvendor model.
782
Maister (1976: 132), Evers (1995: 2, 14f.).
105
and remains the same. The service level increases, however, due to the risk pooling effect.
Perhaps this service level is higher than necessary.
Often the stock-to-demand policy is subject to minimum order and saltus quantity con-
straints as in the case of Papierco. Then the order quantity qi for a specific product at ware-
house i = 1, …, n results from forecast demand minus available inventory di for a specific
period at warehouse i and a constant minimum order quantity m and saltus quantity s. The
saltus quantity demands that only certain multiples are ordered like in Köchel (2007), e. g.
only full pallets.
If a stock-to-demand order policy with minimum order and saltus quantities is followed,
centralization may lead to a reduction in average cycle stock. The order quantity and there-
fore average cycle stock in the decentralized system, where every location places orders for
itself, is more than or equal to the ones in the centralized system, where a common joint
order is placed for all locations783, since often more than forecast demand minus available
inventory is ordered in order to meet minimum order and saltus quantities. In the centra-
lized system the constant minimum order and saltus quantities may be exhausted more.
The order quantity qi for a specific product at warehouse i is
, . (5.1)
The total system wide order quantity in the decentralized system therefore is:
∑ ∑ , . (5.2)
The total system wide order quantity in the centralized system is:
∑
, . (5.3)
The total system wide order quantity in the centralized system is less than or equal to
the one in the decentralized system, because the order quantity function (5.1) is subaddi-
tive:
∑
, ∑ , . (5.4)
For the proof please refer to appendix E.

The centralized ordering effect (COE), the percent reduction of the order quantity by
centralized ordering, is:
783
For Özen et al. (2005: 1) inventory centralization also is a byword for joint ordering: “Groups of retailers
might increase their expected joint profit by inventory centralization, which means that they make a joint
order to satisfy total future demand”.
106
∑
,
1 100 1 100. (5.5)
∑ ,
Centralized ordering can reduce demand and supply uncertainty and thus inventory and
stockouts and increase customer service level by demand pooling, aggregate forecasting
across all locations, and postponement of order allocation to the delivery warehouses. This
will be shown in the following section.
5.3.3.3 Benefits of Centralized Ordering for Papierco

The example of a woodfree white cutsize A-quality copy paper in table 5.2 demonstrates that
the centralized ordering effect for a stock-to-demand order policy with minimum order and
saltus quantities can be significant. This copy paper has the highest annual sales and contribu-
tion margin in warehouse sales of all products at most Papierco locations.
Table 5.2 shows the actual warehouse sales for this copy paper for Papierco's 22 Ger-
man locations and one Dutch location per month in 2007. The sales forecasts were ob-
tained with exponential smoothing. The forecast method was initialized with the 2006 ac-
tual warehouse sales per month per location, i. e. data from 2006 were used to determine
starting values for the exponential smoothing model. The exponential smoothing constant
(column O) that minimized the forecast error (the standard error of the forecast (RMSE))
for the monthly forecasts of 2006 warehouse sales for every location was used for forecast-
ing 2007 sales.784
784
Cf. Ballou (2004b: 299).
107
1 A B C D E F G H I J K L M N O P Q R S
2 Minimum order quantity 22,500 kg Value 0.93 €/kg All values in kg except for the exponential smoothing constant (sm. const.).
3 Saltus order quantity 500 kg Contribution margin 0.23 €/kg Average monthly positive net stock (sum of the positive net stocks divided by twelve).
4
5 Work days 22 20 22 19 20 21 22 23 20 22 22 19
6 Warehouse Location Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Total/Average Sm. Const. MAD BIAS RMSE Safety Stock
7 Aalen
8 Actual sales 43,300 32,608 44,833 26,535 26,935 23,490 27,773 34,888 19,350 36,420 29,615 8,140 353,887 15,835
9 Forecast 27,465 27,624 27,674 27,845 27,832 27,823 27,780 27,780 27,851 27,766 27,852 27,870 333,162 0.01 7,523 1,727 9,922
10 Forecast error 15,835 4,984 17,159 -1,310 -897 -4,333 -7 7,108 -8,501 8,654 1,763 -19,730 20,725
11 Order quantity 27,500 43,500 33,000 45,000 26,500 27,000 23,000 28,000 35,000 22,500 33,000 30,000 374,000
12 Net stock -15,800 -4,908 -16,741 1,724 1,289 4,799 26 -6,862 8,788 -5,132 -1,747 20,113 3,062
13 Sales rate 1,968 1,630 2,038 1,397 1,347 1,119 1,262 1,517 968 1,655 1,346 428
14 Average cycle stock (CS) 8,952 11,868 9,037 14,992 14,757 16,544 13,913 11,380 18,463 13,539 13,154 24,183 14,232
15 Berlin
16 Actual sales 40,165 44,888 37,418 39,375 33,300 41,138 34,318 39,308 41,488 40,243 42,313 8,241 442,195 6,862
17 Forecast 35,059 35,110 35,208 35,230 35,271 35,252 35,311 35,301 35,341 35,402 35,451 35,519 423,454 0.01 6,602 1,562 9,362
18 Forecast error 5,106 9,778 2,210 4,145 -1,971 5,886 -993 4,007 6,147 4,841 6,862 -27,278 18,741
19 Order quantity 35,500 40,000 45,000 37,500 39,500 33,000 41,500 34,000 39,500 41,500 40,500 42,000 469,500
20 Net stock -4,665 -9,553 -1,971 -3,846 2,354 -5,784 1,398 -3,910 -5,898 -4,641 -6,454 27,305 2,588
21 Sales rate 1,826 2,244 1,701 2,072 1,665 1,959 1,560 1,709 2,074 1,829 1,923 434
22 Average cycle stock 15,788 14,097 16,835 16,122 19,004 15,307 18,557 16,019 15,392 15,847 15,323 31,426 17,476
23 Bielefeld
24 Actual sales 44,288 28,438 53,200 25,388 27,913 21,163 22,028 45,038 18,665 40,700 31,663 15,838 374,322 16,381
25 Forecast 28,440 28,599 28,597 28,843 28,809 28,800 28,723 28,657 28,820 28,719 28,839 28,867 344,713 0.01 9,472 2,467 11,780
26 Forecast error 15,848 -161 24,603 -3,455 -896 -7,637 -6,695 16,381 -10,155 11,981 2,824 -13,029 29,609
27 Order quantity 28,500 44,500 28,500 53,500 25,500 28,000 22,500 22,500 43,000 22,500 37,000 31,500 387,500
28 Net stock -15,788 274 -24,426 3,686 1,273 8,110 8,582 -13,956 10,379 -7,821 -2,484 13,178 3,790
29 Sales rate 2,013 1,422 2,418 1,336 1,396 1,008 1,001 1,958 933 1,850 1,439 834
31 Bremen
32 Actual sales 22,150 17,000 22,450 12,913 13,988 17,938 12,663 24,038 11,238 14,188 24,013 16,825 209,404 4,968
33 Forecast 19,224 19,253 19,231 19,263 19,199 19,147 19,135 19,070 19,120 19,041 18,993 19,043 229,721 0.01 4,382 -1,693 4,790
34 Forecast error 2,926 -2,253 3,219 -6,350 -5,211 -1,209 -6,472 4,968 -7,882 -4,853 5,020 -2,218 -20,317
35 Order quantity 22,500 22,500 22,500 22,500 22,500 0 22,500 22,500 22,500 0 22,500 22,500 225,000
36 Net stock 350 5,850 5,900 15,487 23,999 6,061 15,898 14,360 25,622 11,434 9,921 15,596 12,540
37 Sales rate 1,007 850 1,020 680 699 854 576 1,045 562 645 1,092 886
39 Cologne
40 Actual sales 127,123 114,645 241,830 96,180 123,998 111,878 105,668 176,203 84,720 139,303 113,465 97,890 1,532,903 58,566
41 Forecast 105,695 107,838 108,519 121,850 119,283 119,754 118,967 117,637 123,493 119,616 121,585 120,773 1,405,010 0.10 30,095 10,658 45,742
42 Forecast error 21,428 6,807 133,311 -25,670 4,715 -7,876 -13,299 58,566 -38,773 19,687 -8,120 -22,883 127,893
43 Order quantity 106,000 129,000 115,500 255,000 94,000 124,500 111,000 104,000 182,500 80,500 141,500 112,500 1,556,000
44 Net stock -21,123 -6,768 -133,098 25,722 -4,276 8,346 13,678 -58,525 39,255 -19,548 8,487 23,097 9,882
45 Sales rate 5,778 5,732 10,992 5,062 6,200 5,328 4,803 7,661 4,236 6,332 5,158 5,152
47 Dietzenbach (Frankfurt)
48 Actual sales 37,325 44,800 36,765 35,105 28,875 58,725 26,413 82,000 23,588 37,725 34,888 39,073 485,282 22,120
49 Forecast 36,609 36,616 36,698 36,699 36,683 36,605 36,826 36,722 37,174 37,039 37,045 37,024 441,738 0.01 9,555 3,629 15,739
50 Forecast error 716 8,184 67 -1,594 -7,808 22,120 -10,413 45,278 -13,586 686 -2,157 2,049 43,544
51 Order quantity 37,000 37,000 45,000 37,000 35,000 28,500 59,000 26,500 82,500 23,500 37,500 35,000 483,500
52 Net stock -325 -8,125 110 2,005 8,130 -22,095 10,492 -45,008 13,904 -321 2,291 -1,782 3,078
53 Sales rate 1,697 2,240 1,671 1,848 1,444 2,796 1,201 3,565 1,179 1,715 1,586 2,056
55 Dortmund
56 Actual sales 2,238 2,775 8,503 3,800 4,050 6,040 4,225 3,300 3,440 5,190 2,813 2,690 49,064 2,383
57 Forecast 3,625 3,611 3,603 3,652 3,653 3,657 3,681 3,686 3,682 3,680 3,695 3,686 43,911 0.01 1,218 429 1,757
58 Forecast error -1,387 -836 4,900 148 397 2,383 544 -386 -242 1,510 -882 -996 5,153
59 Order quantity 22,500 0 0 0 0 22,500 0 0 0 0 22,500 0 67,500
60 Net stock 20,262 17,487 8,984 5,184 1,134 17,594 13,369 10,069 6,629 1,439 21,126 18,436 11,809
61 Sales rate 102 139 387 200 203 288 192 143 172 236 128 142
63 Ernstroda (Erfurt)
64 Actual sales 6,900 4,888 15,125 7,213 11,788 10,650 14,313 16,263 27,225 18,238 18,525 19,263 170,391
65 Forecast 18,712 12,806 8,847 11,986 9,599 10,694 10,672 12,492 14,378 20,801 19,520 19,022 169,529 0.50 4,756 72 6,254 6,278
66 Forecast error -11,812 -7,918 6,278 -4,773 2,189 -44 3,641 3,771 12,847 -2,563 -995 241 862
67 Order quantity 22,500 0 0 22,500 0 22,500 0 22,500 22,500 23,000 22,500 22,500 180,500
68 Net stock 15,600 10,712 -4,413 10,874 -914 10,936 -3,377 2,860 -1,865 2,897 6,872 10,109 5,905
69 Sales rate 314 244 688 380 589 507 651 707 1,361 829 842 1,014
71 Fellbach (Stuttgart)
72 Actual sales 32,800 37,603 41,795 39,615 39,400 30,535 45,043 49,350 42,663 49,505 52,850 14,850 476,009 11,905
73 Forecast 36,395 36,035 36,192 36,752 37,039 37,275 36,601 37,445 38,636 39,038 40,085 41,361 452,854 0.10 8,071 1,930 10,450
74 Forecast error -3,595 1,568 5,603 2,863 2,361 -6,740 8,442 11,905 4,027 10,467 12,765 -26,511 23,155
75 Order quantity 36,500 32,500 38,000 42,000 40,000 39,500 30,000 46,000 50,500 43,000 50,500 54,500 503,000
76 Net stock 3,700 -1,403 -5,198 -2,813 -2,213 6,752 -8,291 -11,641 -3,804 -10,309 -12,659 26,991 3,120
77 Sales rate 1,491 1,880 1,900 2,085 1,970 1,454 2,047 2,146 2,133 2,250 2,402 782
79 Garching (Munich)
80 Actual sales 56,395 51,943 69,108 46,790 53,673 53,540 56,500 47,680 50,575 59,443 41,563 37,363 624,573 15,463
81 Forecast 40,932 48,664 50,303 59,706 53,248 53,460 53,500 55,000 51,340 50,958 55,200 48,382 620,693 0.50 7,933 323 10,054
82 Forecast error 15,463 3,279 18,805 -12,916 425 80 3,000 -7,320 -765 8,485 -13,637 -11,019 3,880
83 Order quantity 41,000 64,500 53,500 78,500 40,000 54,000 53,500 58,000 44,000 50,500 63,500 35,000 636,000
84 Net stock -15,395 -2,838 -18,446 13,264 -409 51 -2,949 7,371 796 -8,147 13,790 11,427 3,892
85 Sales rate 2,563 2,597 3,141 2,463 2,684 2,550 2,568 2,073 2,529 2,702 1,889 1,966
87 Hemmingen (Hanover)
88 Actual sales 60,453 62,338 75,190 42,813 52,063 47,125 59,500 76,913 41,063 45,175 56,228 36,325 655,186 24,282
89 Forecast 41,620 45,386 48,777 54,059 51,810 51,861 50,914 52,631 57,487 54,202 52,397 53,163 614,307 0.20 13,119 3,407 15,279
90 Forecast error 18,833 16,952 26,413 -11,246 253 -4,736 8,586 24,282 -16,424 -9,027 3,831 -16,838 40,879
91 Order quantity 42,000 64,000 66,000 80,500 40,500 52,000 46,000 61,500 81,500 38,000 43,500 57,000 672,500
92 Net stock -18,453 -16,791 -25,981 11,706 143 5,018 -8,482 -23,895 16,542 9,367 -3,361 17,314 5,008
93 Sales rate 2,748 3,117 3,418 2,253 2,603 2,244 2,705 3,344 2,053 2,053 2,556 1,912
95 Kiel
96 Actual sales 17,728 19,563 26,308 21,080 17,575 24,715 27,015 29,045 21,305 18,883 29,563 14,013 266,793 5,483
97 Forecast 23,673 23,614 23,573 23,601 23,575 23,515 23,527 23,562 23,617 23,594 23,547 23,607 283,006 0.01 4,505 -1,351 5,009
98 Forecast error -5,945 -4,051 2,735 -2,521 -6,000 1,200 3,488 5,483 -2,312 -4,711 6,016 -9,594 -16,213
99 Order quantity 24,000 22,500 22,500 22,500 22,500 22,500 22,500 22,500 25,500 22,500 22,500 24,500 276,500
100 Net stock 6,272 9,209 5,401 6,821 11,746 9,531 5,016 -1,529 2,666 6,283 -780 9,707 6,054
101 Sales rate 806 978 1,196 1,109 879 1,177 1,228 1,263 1,065 858 1,344 738
103 Leizen
104 Actual sales 2,488 2,388 6,025 5,100 4,075 3,700 3,238 4,375 5,613 8,563 7,550 3,125 56,240 2,662
105 Forecast 4,920 4,896 4,871 4,882 4,885 4,876 4,865 4,848 4,844 4,851 4,888 4,915 58,542 0.01 1,611 -192 1,896
106 Forecast error -2,432 -2,508 1,154 218 -810 -1,176 -1,627 -473 769 3,712 2,662 -1,790 -2,302
107 Order quantity 22,500 0 0 0 0 22,500 0 0 0 0 22,500 0 67,500
108 Net stock 20,012 17,624 11,599 6,499 2,424 21,224 17,986 13,611 7,998 -565 14,385 11,260 12,052
109 Sales rate 113 119 274 268 204 176 147 190 281 389 343 164
111 Lohfelden (Kassel)
112 Actual sales 4,650 1,288 4,700 2,763 1,950 388 2,013 2,775 1,963 2,200 2,900 2,400 29,990 2,398
113 Forecast 2,252 2,276 2,266 2,290 2,295 2,292 2,273 2,270 2,275 2,272 2,271 2,277 27,309 0.01 870 223 1,206
114 Forecast error 2,398 -988 2,434 473 -345 -1,904 -260 505 -312 -72 629 123 2,681
115 Order quantity 22,500 0 0 0 0 0 0 0 22,500 0 0 0 45,000
116 Net stock 17,850 16,562 11,862 9,099 7,149 6,761 4,748 1,973 22,510 20,310 17,410 15,010 12,604
117 Sales rate 211 64 214 145 98 18 92 121 98 100 132 126
119 Lunteren (NL)
120 Actual sales 1,513 100 2,625 100 0 1,163 2,150 513 963 2,138 1,863 1,075 14,203 1,414
121 Forecast 437 652 542 959 787 629 736 1,019 918 927 1,169 1,308 10,083 0.20 833 343 988
122 Forecast error 1,076 -552 2,083 -859 -787 534 1,414 -506 45 1,211 694 -233 4,120
123 Order quantity 22,500 0 0 0 0 0 0 0 0 0 0 0 22,500
124 Net stock 20,987 20,887 18,262 18,162 18,162 16,999 14,849 14,336 13,373 11,235 9,372 8,297 15,410
125 Sales rate 69 5 119 5 0 55 98 22 48 97 85 57
108
1 A B C D E F G H I J K L M N O P Q R S
6 Warehouse Location Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Total/Average Sm. Const. MAD BIAS RMSE Safety Stock
127 Mannheim
128 Actual sales 14,475 8,225 21,500 23,625 18,025 18,463 17,050 22,575 12,025 16,913 23,900 11,188 207,964 7,061
129 Forecast 16,620 16,598 16,514 16,564 16,635 16,649 16,667 16,671 16,730 16,683 16,685 16,757 199,773 0.01 4,148 683 4,957
130 Forecast error -2,145 -8,373 4,986 7,061 1,390 1,814 383 5,904 -4,705 230 7,215 -5,569 8,191
131 Order quantity 22,500 22,500 0 22,500 22,500 22,500 22,500 22,500 22,500 0 22,500 22,500 225,000
132 Net stock 8,025 22,300 800 -325 4,150 8,187 13,637 13,562 24,037 7,124 5,724 17,036 10,382
133 Sales rate 658 411 977 1,243 901 879 775 982 601 769 1,086 589
135 Nuremberg
136 Actual sales 2,540 4,488 8,425 9,135 9,488 8,668 10,638 6,400 10,188 7,843 11,113 7,675 96,601 1,598
137 Forecast 10,388 9,603 9,092 9,025 9,036 9,081 9,040 9,200 8,920 9,047 8,926 9,145 110,503 0.10 2,094 -1,158 3,014
138 Forecast error -7,848 -5,115 -667 110 452 -413 1,598 -2,800 1,268 -1,204 2,187 -1,470 -13,902
139 Order quantity 22,500 0 0 22,500 0 0 22,500 0 22,500 0 0 22,500 112,500
140 Net stock 19,960 15,472 7,047 20,412 10,924 2,256 14,118 7,718 20,030 12,187 1,074 15,899 12,258
141 Sales rate 115 224 383 481 474 413 484 278 509 357 505 404
143 Ottendorf-Okrilla (Dresden)
144 Actual sales 1,613 200 650 1,188 625 500 1,200 63 1,188 2,463 963 500 11,153 803
145 Forecast 810 818 812 810 814 812 809 813 805 809 826 827 9,765 0.01 509 116 649
146 Forecast error 803 -618 -162 378 -189 -312 391 -750 383 1,654 137 -327 1,388
147 Order quantity 22,500 0 0 0 0 0 0 0 0 0 0 0 22,500
148 Net stock 20,887 20,687 20,037 18,849 18,224 17,724 16,524 16,461 15,273 12,810 11,847 11,347 16,723
149 Sales rate 73 10 30 63 31 24 55 3 59 112 44 26
151 Queis (Halle)
152 Actual sales 34,400 33,738 34,800 40,538 24,025 28,750 30,938 36,663 25,863 29,650 39,575 23,238 382,178 9,117
153 Forecast 28,771 29,334 29,774 30,277 31,303 30,575 30,393 30,447 31,069 30,548 30,458 31,370 364,321 0.10 5,378 1,488 6,151
154 Forecast error 5,629 4,404 5,026 10,261 -7,278 -1,825 545 6,216 -5,206 -898 9,117 -8,132 17,857
155 Order quantity 29,000 35,000 34,000 35,500 41,500 23,500 28,500 31,000 37,000 25,500 29,500 40,500 390,500
156 Net stock -5,400 -4,138 -4,938 -9,976 7,499 2,249 -189 -5,852 5,285 1,135 -8,940 8,322 2,041
157 Sales rate 1,564 1,687 1,582 2,134 1,201 1,369 1,406 1,594 1,293 1,348 1,799 1,223
159 Reinbek (Hamburg)
160 Actual sales 60,375 50,345 55,478 48,475 64,020 62,413 70,720 81,763 41,658 61,405 62,938 42,615 702,205 16,611
161 Forecast 43,764 48,748 49,227 51,102 50,314 54,426 56,822 60,991 67,223 59,553 60,109 60,958 663,237 0.30 11,003 3,247 13,547
162 Forecast error 16,611 1,597 6,251 -2,627 13,706 7,987 13,898 20,772 -25,565 1,852 2,829 -18,343 38,968
163 Order quantity 44,000 65,500 50,500 57,500 47,500 68,500 64,500 75,000 88,000 34,000 62,000 64,000 721,000
164 Net stock -16,375 -1,220 -6,198 2,827 -13,693 -7,606 -13,826 -20,589 25,753 -1,652 -2,590 18,795 3,948
165 Sales rate 2,744 2,517 2,522 2,551 3,201 2,972 3,215 3,555 2,083 2,791 2,861 2,243
167 Sasbach (Strasbourg)
168 Actual sales 16,790 16,953 18,805 18,165 14,225 15,950 15,005 12,015 7,950 10,223 13,553 12,723 172,357 2,273
169 Forecast 15,090 16,790 16,953 18,805 18,165 14,225 15,950 15,005 12,015 7,950 10,223 13,553 174,724 1.00 2,038 -197 2,390
170 Forecast error 1,700 163 1,852 -640 -3,940 1,725 -945 -2,990 -4,065 2,273 3,330 -830 -2,367
171 Order quantity 22,500 22,500 22,500 22,500 0 22,500 22,500 0 22,500 0 0 22,500 180,000
172 Net stock 5,710 11,257 14,952 19,287 5,062 11,612 19,107 7,092 21,642 11,419 -2,134 7,643 11,232
173 Sales rate 763 848 855 956 711 760 682 522 398 465 616 670
175 Tettnang (Lake Constance)
176 Actual sales 5,238 4,900 6,425 2,238 3,513 4,968 4,235 4,313 4,300 4,988 3,200 3,163 51,481 1,932
177 Forecast 3,306 3,692 3,934 4,432 3,993 3,897 4,111 4,136 4,171 4,197 4,355 4,124 48,349 0.20 1,059 261 1,308
178 Forecast error 1,932 1,208 2,491 -2,194 -480 1,071 124 177 129 791 -1,155 -961 3,132
179 Order quantity 22,500 0 0 0 22,500 0 0 0 0 0 22,500 0 67,500
180 Net stock 17,262 12,362 5,937 3,699 22,686 17,718 13,483 9,170 4,870 -118 19,182 16,019 11,866
181 Sales rate 238 245 292 118 176 237 193 188 215 227 145 166
183 Trierweiler (Trier)
184 Actual sales 19,150 19,728 41,668 19,038 21,400 22,665 27,153 27,925 8,113 30,838 18,025 23,425 279,128 9,166
185 Forecast 21,553 21,529 21,511 21,713 21,686 21,683 21,693 21,748 21,809 21,672 21,764 21,727 260,088 0.01 5,687 1,587 8,062
186 Forecast error -2,403 -1,801 20,157 -2,675 -286 982 5,460 6,177 -13,696 9,166 -3,739 1,698 19,040
187 Order quantity 22,500 22,500 22,500 35,000 22,500 22,500 22,500 23,000 28,000 22,500 22,500 22,500 288,500 SUM(S7:S184) =
188 Net stock 3,350 6,122 -13,046 2,916 4,016 3,851 -802 -5,727 14,160 5,822 10,297 9,372 4,992 245,561
189 Sales rate 870 986 1,894 1,002 1,070 1,079 1,234 1,214 406 1,402 819 1,233
191 Total demand 654,097 603,842 873,626 567,172 594,904 614,565 619,799 823,406 505,144 682,237 663,079 441,638 7,643,509 201,579
192 Aggregate forecast 571,613 588,110 591,256 647,730 631,619 624,276 622,334 621,827 662,143 630,743 641,042 645,449 7,478,140 0.2 95,502 13,781 130,602
193 Forecast error 82,484 15,732 282,370 -80,558 -36,715 -9,711 -2,535 201,579 -156,999 51,494 22,037 -203,811 165,369 43,982
194 Pooled order quantity 572,000 670,500 607,000 930,000 551,000 587,500 613,000 619,000 864,000 473,500 692,500 667,500 7,847,500
195 Non-pooled order quantity 721,000 668,000 599,000 892,000 542,500 636,000 614,500 599,500 872,000 449,500 718,500 661,500 7,974,000
196 Net stock -82,097 -15,439 -282,065 80,763 36,859 9,794 2,995 -201,411 157,445 -51,292 -21,871 203,991 40,987
197 Sales rate 29,732 30,192 39,710 29,851 29,745 29,265 28,173 35,800 25,257 31,011 30,140 23,244
198 Average CS central ordering 251,782 287,217 204,515 364,349 334,311 317,077 312,895 238,270 410,017 292,938 310,619 424,810 312,400
199 Average CS decentral ordering 411,112 430,134 352,512 466,755 428,452 463,209 458,833 369,495 541,441 401,898 446,266 551,405 443,459
200
201 Negative net stock
202 Pooled system 654,175 kg
203 Non-pooled system 845,636 kg
Table 5.2: Effects of Centralized Ordering at Papierco
The order quantity for every month is the demand forecast for that month minus the
net stock (at the end of the period after demand has been satisfied but before the order ar-
rives at the beginning of the new period) left from the previous period plus the quantity to
meet the minimum order and saltus quantity. Supplier lead time is neglected785. If net cycle
stock is negative (the actual sales exceeded the order quantity and the cycle stock on hand),
this demand is either satisfied via transshipments, covered by safety stock786, or backor-
785
Actually a location might run out of stock before the end of the period and have to wait until the begin-
ning of the next period to place and receive a new order. This time might be seen as a replenishment order
lead time.
786
Safety stock is neglected here at first. However, safety stock may also be “defined as the average level of
the net stock just before a replenishment arrives” (Silver et al. 1998: 234, cf. Chopra and Meindl 2007:
305). Here we adhere to the definition “Safety inventory is inventory carried to satisfy demand that ex-
ceeds the amount forecasted for a given period” (Chopra and Meindl 2007: 304).
109
dered. In any case this quantity has to be ordered in addition for the next period to reple-
nish the stock of the location that has transshipped this item or the safety stock or satisfy
the backordered demand. It is assumed that there are no available inventories or backorders
of the copy paper at the beginning of the year 2007. The net stock of the respective period
is the order quantity minus the actual sales plus the previous period's net stock. If the fore-
cast of next period's sales can be met with net stock no order is placed with the supplier
(the order quantity is zero). This occurs at the locations Bremen, Dortmund, Ernstroda,
Leizen, Lohfelden, Lunteren, Mannheim, Nuremberg, Ottendorf, Sasbach, and Tettnang,
where the monthly forecasts and actual sales are lower than the minimum order quantity
(except in Bremen in August and November and Ernstroda in September).
For central ordering (total demand) the monthly actual sales are the sum of the individ-
ual locations' monthly sales (row 191). The total 2007 order quantity of this copy paper
with central ordering for all locations (pooled order quantity) 7,847,500 kg (cell N194) is
lower than the sum of the total order quantities of the individual locations ordering sepa-
rately (non-pooled order quantity) 7,974,000 kg (cell N195). The centralized ordering
effect (equation (5.5)) amounts to 1.59 %. With central ordering Papierco could have saved
ordering 126,500 kg, 117,645 € (= 126,500 kg × 0.93 €/kg) purchasing costs, as well as
average cycle inventory and the associated inventory carrying cost just for this one product
by improved exhausting of the minimum order and saltus quantity.
For some months the non-pooled order quantity is lower than the pooled one. This is
evoked by some locations not placing an order in some months and serving demand from a
minimum order quantity and net stock, as the forecasts and actual sales are much lower
than the minimum order quantity. However, in these cases higher inventory carrying costs
have to be borne, as the ordered minimum order quantity minus sales remains in stock for
several periods. Negative net stock also postpones order quantities to later periods. Be-
sides, a higher order quantity than the forecast to meet minimum order and saltus quantities
works as a buffer for positive forecast errors (if the actual demand exceeds the forecast):
The absolute value of the negative net stock is always less than the forecast error in these
cases. Nevertheless, on the whole a lower total order quantity is needed to serve the same
total annual demand with central than with decentral ordering.
We approximated the average cycle stock per warehouse location as follows: We calculated
the sales rate (actual sales divided by work days). Sales can only occur on work days. We de-
termined the work days per month for the year 2007 (row 5). In case the work days in the dif-
ferent German federal states differed, we took the maximum. We then determined the inventory
110
level at the end of each working day after sales, deducting the sales rate from the previous end
of day inventory. Whenever the end of day inventory was negative we replaced it with zero
physical inventory. The inventory level at the beginning of the month is the order quantity plus
the previous month's end inventory level. We thus assumed that backordered demand (negative
end of month net stock) is satisfied instantaneously at the beginning of the new month when
the order quantity arrives. Finally we determined the arithmetic mean of the end of work
day inventory levels as the average cycle inventory. This also confirms that today average
inventory at Papierco is so high relative to actual sales partly due to minimum order and
saltus quantities.
The total average cycle stock with central ordering (row 198) is lower than with decen-
tral ordering (row 199) in every month and for the whole year 2007, because of improved
exhausting of minimum order and saltus quantities. For this copy paper Papierco could
have saved 131,059 kg of average cycle stock and the corresponding inventory carrying
costs of 30,471 € (= 131,059 kg × 0.93 €/kg × 0.25) in 2007.
The sum of the average monthly positive net stock of the individual locations
(184,234 kg) in column N is larger than the average monthly positive net stock with central
ordering in cell N196 (40,987 kg).
The sum of the forecast error statistics mean absolute deviation (MAD 142,459), bias
(BIAS 29,560), and root mean square error (RMSE 190,306) for the individual locations in
columns P, Q, and R is higher than the error statistics for the central order (95,502, 13,781,
and 130,602 respectively) in cells P, Q, and R192.787
This confirms788 that the aggregate forecast of monthly sales of this copy paper for all
23 considered locations is more accurate than the disaggregate forecast for the individual
locations. This leads to a higher inventory availability, i. e. a smaller total stockout or
backorder quantity (negative net stock) and thus smaller stockout or backorder costs in the
system with central ordering (654,175 kg) than in the system with decentral ordering
(845,636 kg), if safety stock is neglected.
According to the One-Sample Kolmogorov-Smirnov Test, the H0-hypothesis that the
forecast error at the locations Aalen, Berlin, Bielefeld, Bremen, Cologne, Dietzenbach,
Dortmund, Ernstroda, Fellbach, Garching, Hemmingen, Kiel, Leizen, Lohfelden, Lunteren,
Mannheim, Nuremberg, Ottendorf, Queis, Reinbek, Sasbach, Tettnang, and Trierweiler
787
“MAD is […] the average of the absolute differences between the actual […] and the forecast” sales.
“BIAS is the average of the differences between actual and forecast” sales. “RMSE is [the] square root of
the average of the squared differences between the actual and the forecast” sales (Ballou 2004c: 5f.).
788
Cf. footnote 456.
111
and of total demand is normally distributed cannot be rejected with an asymptotic signific-
ance (two-tailed) p of 0.984, 0.317, 0.879, 0.872, 0.485, 0.268, 0.582, 0.996, 0.513, 0.974,
0.970, 0.922, 0.973, 0.695, 0.966, 0.984, 0.681, 0.864, 0.832, 0.711, 0.758, 0.991, 0.851,
and 0.968 respectively. The One-Sample Kolmogorov-Smirnov Test for the exponential
distribution shows a higher asymptotic significance (two-tailed) for Cologne (p = 0.820),
Dortmund (p = 0.936), Fellbach (p = 0.824), Leizen (p = 0.993), Lohfelden (p = 0.911),
Nuremberg (p = 0.839), Reinbek (p = 0.830), Trierweiler (p = 0.977), and total demand
(p = 0.993). However, for Fellbach and Reinbek there are 3, for Cologne, Dortmund, Loh-
felden, Trierweiler, and total demand 6, for Leizen and Nuremberg 7 values outside the
specified distribution range. These values are skipped. Kiel shows a higher asymptotic sig-
nificance (two-tailed) of 0.958 in the One-Sample Kolmogorov-Smirnov Test for the uni-
form distribution. As Poisson variables are non-negative integers and negative values occur
in the data, the One-Sample Kolmogorov-Smirnov Test cannot be performed for the Pois-
son distribution. In conclusion, the forecast error can be assumed to be rather normally
distributed at the different locations. According to Thonemann (2005: 258) most applica-
tions assumed a normally-distributed forecast error.
Nevertheless, we use the empirical distribution function of the forecast error to deter-
mine the safety stock. The empirical distribution function F(y) gives the probability that
the forecast error is less than or equal to y, i. e. F(y) = P (forecast error Y ≤ y). In order to
have an inventory availability of 91.7 %, we determine the y for P = 91.7 %. If we carry y
as safety stock, this will cover demand that exceeds the forecast with a probability of
91.7 %, as the forecast error is less than or equal to this safety stock with the probability of
91.7 %. The safety stock that ensures a service level of 91.7 % is shown in column S for
every location. The sum of the individual locations' safety stocks (245,561 kg) in cell S188
is larger than the safety stock with central forecasting and ordering for all locations
(201,579 kg) in cell S191, as the aggregate forecast is more accurate than the disaggregate
one. Papierco could have saved 43,982 kg of safety stock and the associated inventory
holding cost of 10,226 € (= 43,982 kg × 0.93 €/kg × 0.25) just for this copy paper in 2007.
Although Papierco forecasts demand and therefore should use safety stock as protection
against the forecast error and not demand uncertainty789, it uses the average past demand
for the copy paper to determine safety stock.
789
Cf. Thonemann (2005: 255f.).
112
With decentralized ordering there would have been 123,215 kg negative net stock that
could not have been absorbed by the safety stock for all locations in 2007. We arrived at
this figure by adding the negative net (cycle) stock and safety stock for every location for
every month, whenever the absolute value of the former was greater than the one of the
latter, and then adding up all the negative sums. With centralized ordering 80,486 kg of
demand would have exceeded available inventory (cycle stock and safety stock). In the
decentralized system the 123,215 kg of demand could have been satisfied by transship-
ments from other warehouses, backordered, or lost. In the centralized system, transship-
ments would not have been possible, as demand would have exceeded the order quantity
and safety stock for all locations, and demand could have been only backordered or lost.
Therefore one could argue that in this regard the service level in the decentralized system
would have been higher. Transshipping 123,215 kg of this copy paper in the decentralized
system would have cost at least 8,561 € = 123.215 t × 69.48 €/t (cf. section 5.3.2.3). Losing
the contribution margin of selling the 80,486 kg of this copy paper would have cost
80,486 kg × 0.23 €/kg = 18,512 € in the centralized system. Thus the net stockout costs
(lost contribution margin in the centralized system minus the transshipment cost in the de-
centralized one) amount to 9,951 € = 18,512 € - 8,561 € at the most. If all the unsatisfied
demand had been backordered in both systems, backorder costs would have been higher in
the decentralized than in the centralized system because of the backordered volume.
Overall, Papierco could have saved at least 30,746 € in inventory holding costs just
for this one copy paper in 2007, if it had used centralized planning and ordering instead of
the current decentralized one: 30,746 € = 30,471 (saved inventory holding costs for aver-
age cycle stock) + 10,226 € (saved inventory holding costs for safety stock) - 9,951 € (net
stockout costs).
This example highlights that centralized ordering enables to reduce order quantities and
thus average cycle inventory and inventory carrying costs by exhausting minimum order
and saltus quantities better as well as to reduce safety stock and improve overall customer
service level (inventory availability) or reduce negative net stock as the aggregate forecast
is more accurate than the disaggregate one.
Therefore, we propose a central distribution requirements planning with aggregate
forecasting and central ordering for all Papierco locations. After placing a joint order, cen-
tral planning would receive a notification of dispatch from the producer when the ordered
products are produced and the trucks are ready to deliver them. Then Papierco tells the
supplier which warehouses to deliver to according to current demand. Full truck loads can
113
be delivered to single warehouses or one truck can deliver to several warehouses on a deli-
very route. The delivery destination decision is postponed until the ordered products are
produced and ready to be delivered from the production facility to the delivery warehouses
(geographic or logistics postponement).
Today each warehouse location forecasts its warehouse sales and places orders with the
suppliers individually. The warehouse receives a dispatch notification from the supplier
before the delivery truck leaves the production facility to deliver the ordered products to
the warehouse. The delivery destination is determined at the time of placing the order.
With today's decentralized release orders within the centrally negotiated framework
contracts every Papierco location orders all products for itself.
With centralized ordering each location or company of the eight ones comprising Pa-
pierco could process the orders for one part of the product line, e. g. a main product group,
for all considered 23 locations. This would lead to the following advantages:
 Fewer orders at the suppliers (1/23 of today's decentralized orders), lower order
costs, improved utilization of minimum order and saltus quantities, lower average
inventory, higher turnover, lower capital commitment, lower inventory holding
costs.
 Only one procurement manager per product line part or product, higher specialist
knowledge about this part of the assortment, improved forecasting, improved
matching of supply and demand, improved potential utilization (negotiation power),
closer collaboration with suppliers, long-term planning with suppliers, cost degres-
sion potentials for Papierco and its suppliers, reduced lead times.
 Directing supplier deliveries to certain warehouse locations at short notice (higher
flexibility, product availability, and customer service level, reduced transshipments
and inventories).
This approach would maintain the importance of the legally independent companies
forming Papierco and their decentral locations. The employees of the purchasing depart-
ments do not have to relocate to a physically central ordering department, but can remain at
their locations. They may become more motivated by their increased responsibility, as they
do not place replenishment orders for their own location only anymore, but for all Papierco
locations albeit just for a certain part of the product line. Thus possible disadvantages of
centralized ordering such as decreased responsiveness and sales due to lacking local know-
114
ledge790 can be mitigated as well. Procurement employees may have more time for im-
proved forecasting and collaboration with suppliers and other departments or their number
may be reduced.
The number of the suppliers' trips to the warehouses would probably remain unchanged.
Perhaps the supplier would charge a fee for delivery routes, if a truck load is to be trans-
ported to more than one Papierco location. This charge is unlikely to outweigh all the other
stated benefits. The products, however, have to be managed more intensively by procure-
ment managers. Operations and rules in all locations have to be standardized. Necessary
data have to be available and correct.
Orders would still be placed according to order suggestions generated by DRP consis-
tent with customer orders, demand forecasts, inventory levels, and supplier lead time. Of
course the ERP software has to be adapted to plan across all Papierco locations.
The paper industry shows long production lead times, expensive holding costs for in-
ventory surplus, and highly random demand at the retailers. Orders can be backlogged only
for a few days at the most in case of a stockout or the sale is lost. These conditions favor
centralized ordering according to Eppen and Schrage (1981: 51) and Schoenmeyr (2005:
5). Total demand is less variable than demand at the individual stores, which makes con-
solidated distribution most effective according to Cachon and Terwiesch (2009: 340). The
time from placing an order with the supplier until production of the order is finished is
longer than the delivery time from the production facility to the delivery warehouses. This
corresponds to Cachon and Terwiesch's (2009: 340) observation that consolidated distribu-
tion “is most effective […] if the lead time before the distribution center is much longer
than the lead time after the distribution center”. Replenishment coordination is most bene-
ficial for high fixed order/transportation cost, large truck capacities, and many retailers.791
The minimum order and saltus quantity restrictions Papierco has to observe can be sub-
sumed under large fixed order or transportation cost and truck capacities, and Papierco has
many delivery warehouses in Germany.
In contrast to Eppen and Schrage (1981), Gürbüz et al. (2007: 305), and Cachon and Ter-
wiesch (2009) our considerations do not contain a distribution center (here we follow
Schonmeyr's (2005: 4) proposition), Papierco uses a stock-to-demand order policy with min-
imum order and saltus quantities and transshipments between regional warehouses, weekly
demand often does not follow a theoretical distribution (cf. section 5.3.3.1), demand at the
790
791
Gürbüz et al. (2007: 305).
115
different warehouses is correlated, and supplier lead time is largely neglected. Our centra-
lized ordering model does not increase the distance a unit must travel from the supplier to the
delivery warehouses, unless it is delivered on a delivery route with several stops.
Eppen and Schrage (1981) consider centralized ordering policies in a periodic-review
base-stock multi-echelon, multi-period system in the conglomerate for steel industry with
independent normally distributed stationary random demands and identical proportional
costs of holding and backordering at N warehouses, and no transshipments between the
warehouses. In Eppen and Schrage (1981) products are ordered and quantity discounts are
possible. Papierco just calls quantities of products from the supplier within the bounds of
previously centrally negotiated skeleton agreements. Possible quantity discounts have been
granted before.
Diruf (2005, 2007) and Heil (2006: 169) examine only the demand-pooling savings of
postponement, centralized ordering, and lateral transshipments between cooperating possi-
bly independent fashion retailers with normal and lognormal792 demand, equally big sales
areas793, and a newsboy structure. In their model the central ordering system leads to lower
costs than the local ordering system with transshipments unless over- and underage costs
are equal. In this case both systems achieve the same improvements compared to a system
without risk pooling.794
Hu et al. (2005) deal with the impact of emergency transshipments on (s, S) (order-up-
to) ordering policies and centralized ordering.
Gürbüz et al. (2007: 296, 305) consider only one outside supplier, but inventory and
transportation costs and four order-up-to policies.
Cachon and Terwiesch (2009) consider consolidated distribution in retail with 100
stores, an order-up-to-model, Poisson demand at the retail stores per week795, normally-
distributed demand at the retail distribution center796, lead times, and backordering797.
One could still determine an optimal delivery allocation procedure and order policy for
Papierco, but this would go beyond the scope of this thesis. The reader is referred to e. g.
Silver et al. (1998) for a detailed treatment of inventory control policies.
792
Heil (2006: 145f.).
793
Heil (2006: 169).
794
Diruf (2005: 59f.), Heil (2006: 110).
795
796
797
116
Alfaro and Corbett (2006: 24) find that within certain quite wide ranges of suboptimali-
ty of inventory control policies, risk pooling is better than optimizing the policy. This
range expands with the number of SKUs. Papierco carries thousands of SKUs.
Harrison and Skipworth (2008: 193) also find that form postponement in three companies
improved responsiveness of manufacturing, although it had not been implemented theoreti-
cally ideally.
Finally, risk pooling reduces decision errors, biases, and local suboptimization of boun-
dedly rational decision makers and thus total costs “by pooling decision errors across loca-
tions”, even if demands are perfectly positively correlated or deterministic (behavioral
benefits of inventory pooling)798.
Therefore, introducing risk pooling methods at Papierco at first, does not hurt.
5.3.4 Product Pooling
For the last years more and more products have been added to Papierco's product line without
dropping any other ones. The product range meanwhile comprises around 20,000 current
products, some with few sales, and is planned to be diversified further especially in the busi-
ness units printing accessories and screens and signs. This diversification leads to higher in-
ventories, as every warehouse has to store at least some units of every product in order to be
able to deliver every product specification, or more transshipments. Product availability de-
creases as forecasts for more and more individual items are less accurate.
“Every firm wishes to be ʻcustomer focusedʼ or ʻcustomer oriented,ʼ which suggests
that a firm should develop products to meet all of the needs of its potential customers. Tru-
ly innovative new products that add to a firm's customer base should be incorporated into a
firm's product assortment. But if extra product variety merely divides a fixed customer
base into smaller pieces, then the demand–supply mismatch cost for each product will in-
crease. Given that some of the demand–supply mismatch costs are indirect (e.g., loss of
goodwill due to poor service), a firm might not even realize the additional costs it bears
due to product proliferation”799.
Already in 1987 Röttgen remarked due to the cut-throat competition fine paper whole-
salers are forced to distinguish themselves from their competitors. This is possible by in-
tensifying some selected distribution functions rather than accepting additional ones. Thus
paper wholesalers seek to diversify their product line more and more to exhaust the cus-
798
Su (2008: 33).
799
117
tomer potential as much as possible.800 Fine paper wholesale has to abandon the idea to be
a specialist for all paper grades. This implies prioritizing within the product range instead
of a sprawling product line diversification. An escalating diversification will lead to dis-
proportionately high costs and continue to keep the return low.801
According to the FPT-Workshop No. 4 in Munich on October 24, 2007 the variety of
paper grades is narrower in the U.S. and is not demanded by customers (printing offices) in
Germany either. The majority of printing offices deny the necessity of the present variety
of grades. For them operativeness, color consumption, quality consistency, and mixability
of paper grades are decisive and not a new brand or grade.802
A FPT study supports this: A change in demand behavior for paper grades seems to gain
momentum away from the unclear vast variety of grades towards a homogenization. The
printing industry foremost demands good operability and printability of the paper in the
printing machine. Paper grades with standard functionality (reproducible operability and
printability) and multifunctionality are the focus of the forthcoming development.803
Furthermore a too large product variety might backfire, as customers might not like to
choose from many alternatives.804 Toyota's and Nissan's fast introduction of a series of
products, for instance, was counterproductive as it confused customers.805
For all these reasons Papierco should rationalize its product line by product pooling
taking into account its business strategy, demand and buying interdependencies between
products, customer needs, and cooperation with suppliers806 also in order to simplify its
procurement planning. Serving “demand with fewer products” reduces “demand variabili-
ty, which leads to better performance in terms of matching supply and demand (e.g., higher
profit or lower inventory for a targeted service level)”807.
5.3.5 Inventory Pooling
Nowadays there are bilateral agreements between some Papierco member companies to store
slow-moving items for one another. However, there is no transparency, which articles are
stored at which location, and the partly sought selective stock keeping is not pursued strategi-
cally and economically.
800
Röttgen (1987: 83).
801
Röttgen (1987: 124).
802
FPT (2007b).
803
FPT (2007a).
804
Chain Drug Review (2009a), Hamstra (2009), Ryan (2009).
805
Pine II et al. (1993), Lau (1995).
806
Kulpa (2001), Hamstra (2009), MMR (2009), Pinto (2009a, 2009b), Thayer (2009: 23).
807
118
If a new product is launched, it is stored at only three locations first. If it has been sell-
ing well for half a year, further locations start inventorying it and it may be stocked at
every Papierco location in the end.
Papierco's uniform price list suggests immediate availability of all paper grades at every
warehouse location. Due to the increasing range of products, however, availability of every
product at every location cannot be assured anymore. Some products are only available at
one warehouse in Germany. Thus demands for transshipments rise and the 24-hour delivery
service cannot be guaranteed anymore. Transshipment errors increase, especially if a product
is transshipped via other warehouses. Often truck capacity is exceeded. Papierco's competi-
tive advantage of delivering any paper grade mostly within 24 hours is fading.
The product master record contains 61,132 different products. Thereof merely 23,740
ones show stock on hand at at least one warehouse. Of these only 9,868, of the 61,132 only
14,565 ones, are planned automatically with DRP. The product master record should be rid
of inactive products with no sales.
Product purchase planning and storing is distributed unequally between the eight Pa-
pierco member companies as figures 5.8 and 5.9 show. If it was distributed more evenly
at least for fast movers, product availability could be increased and transshipments de-
creased (cf. section 5.3.2.3).
Along the same line, Pasin et al. (2002) simulate pooling of equipment for homecare
service for a group of local community service centers (CLSCs) in the Montreal, Canada
region. Overall pooling reduces shortages (rental costs) with the same inventory level, but
the larger CLSCs with considerable equipment overcapacity may lose, because they almost
never need to borrow equipment from other CLSCs, but have to bear the disadvantages
(wear-and-tear costs). A more even distribution of over-capacity reached by inter-CLSC
sales or an equivalent financial compensation is much more effective than minimum stock,
maximum contribution or maximum debt rules in producing an overall reduction in costs
without penalizing any of the partners in the pooling process.
We would have liked to design an optimal policy of strategic stock-keeping (inventory
pooling through selective stock keeping). However, the necessary data such as the turnover
or (average) inventory per SKU were not available.
119
7000
6057
6000
5000 4879
4624
4261 4197
Number of Products
3872 3936
4000
3124
3003
3000
2651
2000
1668
926 959 939

1000
703
296
0
100 200 300 400 500 600 700 800
Papierco Member Company (Code)
Number of products planned with DRP (planning code = 1)

Number of products not planned with DRP (planning code = 0), but with stock on hand
Figure 5.8a: Product Purchase Planning at Papierco's Member Companies
8000
7537
7000
6000
Number of Planned Identical Products
5000
4000
3000
2229
2000
1487
992
1000 735
639 656
288
0
1 2 3 4 5 6 7 8
Number of Papierco Member Companies
Figure 5.8b: Product Purchase Planning at Papierco's Member Companies
120
14000
12145
11083
12000
10000
8664
8596
8446
Number of SKUs
7309
7303
8000
Inventory (t)
6555
6447
6208
6152
6144
5960
5407
5339
6000
5077
5029
5021
4974
4970
4970
4853
4835
4835
4831
4464
4459
4033
4029
4025
3915
3791
3567
3424
3274
3223
4000
3036
2982
2972
2972
2367
1983
1687
1357
1131
2000
848
840
782
492
443
418
322
295
96
50
11
1
0
0
0
0
0
0
0
101 102 103 201 202 203 301 302 401 402 501 502 503 601 602 603 701 702 703 704 801
Warehouse
Number of routinely stocked SKUs (product code P = purchase including raw materials) Number of SKUs with stock on hand Inventory (t)
Figure 5.9: Stockkeeping at Papierco's Warehouses in February 2008
5.3.6 Challenges in IT and Organization

Today too little process-oriented information exchange between departments inhibits business
processes. Data are not transparent enough. If they are recorded at all, often they can only be
obtained from the diverse, little integrated electronic data processing (EDP) systems with great
effort. Too many subsystems and interfaces result in data transfer problems, too much data
maintenance, and high costs. A lot of workarounds were programmed in the standard ERP
system for Papierco, as many locations wanted to keep some unique business processes. How-
ever, the standard ERP system should standardize the business processes or one does not de-
rive the full benefit from the standard software.808 CLM (1995: 6) also concludes that “the vast
majority of available technology capable of facilitating process integration is underutilized”.
Papierco's different data analysis tools often show different numerical results for the
same data query specifications. Data are often not available on a disaggregate individual
product basis, but only for product groups. Analyses are hampered by different product
codes for the same product for drop-shipping and warehouse sales, different periods (days,
weeks, and months), units (sheets and kilograms), and product groupings and levels in dif-
808
Fleisch et al. (2004), Laudon and Laudon (2010: 369).
121
ferent parts of the EDP system, too little knowledge about the analysis tools at Papierco's
locations, as well as a bad data structure and maintenance. The delivery situation is not
transparent enough: There are customer and delivery addresses. The delivery addresses are
not assigned to customers uniquely. Sales can only be attributed to customer and not to
delivery addresses. Some data are not recorded at all, such as cycle, safety, and average
stock, as well as turnover per product, utilization of safety stock, cycle time, actual supplier
lead times, and actual transshipment costs. Some computerized reports give wrong (calcu-
lated, not actual) safety stock levels for product groups, so that often the shown safety
stock is higher than the actual total inventory on hand. Certain costs can only be obtained
by checking every single cost unit. Higher cost awareness might lead to higher return. Set-
ting up a systematic controlling might be worthwhile.
Data and information are passed on by employees only hesitantly and after direct re-
quest. Important extra information to get a complete idea often is omitted. This might be
interpreted as a defense strategy to secure the job and power position by exclusive know-
ledge. Any criticism can be invalidated by pointing out that certain information had not
been considered. Similar “micropolitical games” were observed e. g. by Reihlen (1997: 3)
in a heating technology company. Change in logistical practices often faces resistance.809
This can be remedied with a company culture that promotes transparency.
Papierco consists of eight different and legally independent companies, but is and
wants to be seen as a single uniform company. However, employees rather feel obliged to
their individual member companies that pay their salaries. This sometimes leads a member
company to transship products for its own warehouse locations first and leave the goods
bound for other member companies' locations behind. Products ought to be transshipped
according to urgency and not company affiliation. Papierco should not optimize locally but
globally.
Logistics is seen very restrictively within Papierco as storing, handling, and transport-
ing goods. Taking on its interdepartmental function, logistics should collaborate closely
with and support management, sales, purchasing, finance, accounting, and controlling.
During this study differences in managing Papierco's different locations became obvious.
Papierco should standardize its operational policy, measure performance uniformly, and re-
structure its distribution system cautiously. For this a consistent and faultless data mainten-
ance and accessibility is essential. Some data are deficient for logistical management.
809
CLM (1995: v, 6), Ballou et al. (2004b: 550), Graman and Magazine (2006: 1077).
122
Most of these challenges are common in a lot of companies. Papierco is addressing

them successfully and is a profitable company.
5.4 Summary
The paper wholesale company Papierco can increase its competitiveness and reduce customer
demand and supplier lead time uncertainty by the risk pooling methods transshipments, prod-
uct substitution, postponement, centralized ordering, product pooling, and inventory pooling.
The proposed optimization of the customer allocation to the individual warehouses can
lead to considerable transportation cost savings near-term and without much investment
burden. Customers should be served by the warehouse nearest to them.
The following suggestions can be made to reduce inventory and transshipments, which
have increased for the last years, and maintain or improve the customer service level and
profitability:
(1) The organizational structure and logistics should support the business
processes as well as possible.
(2) The company culture should promote transparency and a process view.
(3) Improve the fragmented, little integrated, little standardized, not process- but
function-oriented system landscape. This would also lower the disproportio-
nately high IT costs.
(4) Necessary data should be collectable faultlessly.
(5) The demand planning810 and purchasing policy should be improved, as this
will result in higher product availability (customer service level) and less
transshipments and inventory.
(6) Create a uniform cross-functional demand planning process, in which sales,
product managers, and purchasing collaborate to contain extraordinary de-
mands through the sales assistants' and product managers' profound market
knowledge and enable a better demand forecast and thus order policy.
(7) Communicate and coordinate inventory reductions at the different member
companies and warehouses. If a company is member of a purchasing group, it
cannot act autonomously anymore.
810
Methods 11 (Exponential Smoothing), 6 (Least Squares Regression), and 9 (Weighted Moving Average)
should be admitted to the DRP forecast run besides today's exclusively used methods 4 (Moving Aver-
age) and 10 (Linear Smoothing), as they gave good results. The forecast should be based on the last 12 to
24 months of sales history, if system performance permits this.
123
(8) Establish (not necessarily physically) centralized procurement, which fore-

casts and orders for all of Papierco's German locations on the product level
and directs the suppliers' deliveries to the warehouses according to current
demand (consolidated distribution with pooling over the outside-supplier lead
time).
(9) Foster a close, reliable, long-term collaboration and exchange of information
between suppliers, Papierco, customers, and individual departments.
(10) Safety stock and transshipments can be reduced without hurting product avail-
ability by decreasing lead time (uncertainty) through improved collaboration
with suppliers and reducing demand uncertainty by better forecasts and infor-
mation gathering and usage.
(11) Determine safety stocks more definedly according to the target customer ser-
vice level, forecast error, supplier lead time, and lead time variability.
(12) Enable determining the correct inventory levels per product and utilization of
safety stock.
(13) Rationalize the product line considering the business strategy, customer needs,
product interdependencies, and cooperation with suppliers (product pooling).
(14) Establish transparency regarding the storage locations of products.
(15) Pursue selective stock keeping (inventory pooling) strategically economically.
(16) Use alternatives to transshipments (product substitution and form postpone-
ment).
(17) Papierco's emergency transshipments are economically worthwhile on aver-
age.
(18) However, they should only be used, if the contribution margin of a sale is at
least 69.48 €/t.
(19) Establish transshipment cost rates that reflect the true costs.
(20) Set up clear rules for the sales assistants for using transshipments.
(21) Transshipment costs should be made transparent to the sales assistants and
perhaps be deducted from the premium-determining contribution margin.
We would have liked to always evaluate the cost savings attainable by these suggestions for
improvement in money units, but necessary cost data and information were not provided. A
consultancy described similar problems of obtaining data at Papierco.
124
Human resource factors, transaction costs, and local suboptimization seem to be impor-
tant in risk pooling and transshipments. These factors are commonly neglected in the risk
pooling literature.
This application shows that risk pooling can be used effectively as a guiding umbrella
principle in systematically restructuring a distribution system to cope with supplier lead
time, customer demand, and forecast uncertainty, as well as product variety and reduce
costs and inventory, increase customer service, and make a company more competitive.
Risk pooling methods can be used side by side to take advantage of their benefits, while
making up for disadvantages of other ones.
Our Risk Pooling Decision Support Tool is suitable to determine appropriate risk pool-
ing methods for a company that faces lead time and demand uncertainty. Some of the me-
thods determined with the help of this tool are already applied by Papierco nowadays
(transshipments, product substitution, inventory pooling, and postponement), but can and
should be improved. The other ones still make sense after a more thorough investigation
(centralized ordering and product pooling).
Centralized ordering can reduce average cycle inventory, even if a stock-to-demand or-
der policy is followed and minimum order and saltus quantities have to be observed. It also
enables to decrease safety stock and improve inventory availability, as the aggregate fore-
cast for the central order is more accurate than the disaggregate one for the decentral or-
ders.
125
6 Conclusion
Our main novel contributions lie in the (1) comprehensive and concise definition of risk pool-
ing distinguishing between variability, uncertainty, and risk, (2) holistic review and value-
chain-oriented structuring of the plethora of research on risk pooling methods in business lo-
gistics, especially on inventory pooling, according to their uncertainty reduction abilities,
(3) developing tools to help in comparing and choosing appropriate risk pooling methods and
models for different economic conditions based on a contingency approach, (4) applying these
tools to German paper wholesale, and (5) conducting a survey on the recognition and usage of
the various risk pooling concepts and their associations in 102 German trading and manufac-
turing companies.
Risk pooling in business logistics can reduce total variability of demand and/or lead
time and thus uncertainty and the possibility of not achieving business objectives (risk) by
consolidating individual variabilities (measured with the standard deviation) of demand
and/or lead time. These individual variabilities are consolidated by aggregating demands
(demand pooling) and/or lead times (lead time pooling).
This reduction in uncertainty allows to reduce inventory without reducing the custom-
er service level (product availability) or to increase the service level without increasing the
inventory or a combination of both and to cope with product variety. Risk pooling is ex-
plained by the Minkowski-inequality, the subadditivity property of the square root of non-
negative real numbers, and the balancing effect of higher-than-average and lower-than-
average values of a random variable.
Risk pooling usually shows increasing returns, but diminishing marginal returns. Its
benefit generally increases with decreasing correlation of pooled demands and/or lead
times and concentration of uncertainty as well as increasing variability.
Risk pooling in business logistics comprises well over 600 publications mostly contain-
ing modeling research. After the concept borrowed from modern portfolio and insurance
theory has been mainly applied to inventory pooling, meanwhile research focuses on post-
ponement, transshipments, and the coordination of risk pooling arrangements and cost and
profit allocation through contracts. The researchers are of different backgrounds, orienta-
tion, and prominence, so that they use inconsistent terms, frameworks, and structures and
make our thorough literature review, structuring, and definitions essential.
We identified ten risk pooling methods: (1) inventory pooling, (2) virtual pooling,
(3) transshipments, (4) centralized ordering, (5) order splitting, (6) component commonali-
126
6 Conclusion
ty, (7) postponement, (8) capacity pooling, (9) product pooling, and (10) product substitu-
tion.
These methods can be implemented everywhere along the supply chain and mainly per-
tain to the value activities storage (inventory pooling), transportation (virtual pooling and
transshipments), procurement (centralized ordering and order splitting), production of
goods and services (component commonality, postponement, and capacity pooling), and
sales (product pooling and product substitution). We presented a classification of risk
pooling methods according to their ability to pool demands and/or lead times.
They all but order splitting can reduce demand uncertainty, order splitting only lead
time uncertainty, capacity pooling, component commonality, inventory pooling, product
pooling, and product substitution only demand uncertainty. Transshipments, virtual pool-
ing, postponement, and central ordering may dampen both uncertainties.
(1) Inventory pooling is the consolidation of inventories, e. g. by inventory or ware-
house system centralization or selective stock keeping, in order to reduce inventory hold-
ing and shortage costs through risk pooling. Its risk pooling effect in terms of inventory
savings can be quantified with the square root law (SRL), portfolio effect (PE), and in-
ventory turnover curve.
The SRL states that the total system wide stock of n decentralized warehouses is equal
to that of a single centralized one multiplied with the square root of the number of ware-
houses n. Despite confusion in the literature it applies to regular stock, if an economic
order quantity (EOQ) order policy is followed, the fixed cost per order and the per unit
stock holding cost, demand at every decentralized location, and total system demand is the
same both before and after centralization. It is valid for safety stock, if demands at the
decentralized locations are uncorrelated, the variability (standard deviation) of demand at
each decentralized location, the safety factor, and average lead time are the same at all lo-
cations both before and after consolidation, average total system demand remains the same
after consolidation, no transshipments occur, lead times and demands are independent and
identically distributed random variables and independent of each other, the lead time va-
riance is zero, and the safety-factor approach is used to set safety stock for all facilities
both before and after consolidation. It applies to total stock (the sum of regular and safety
stock), if the assumptions stated above of the SRL both as applied to regular and safety
stock hold (cf. appendix B). It does not hold, if its assumptions are not fulfilled.
If an EOQ policy is followed and constant fixed costs per order, holding costs, and total
demand are assumed for all locations before and after centralization, the total order fixed
127
6 Conclusion
cost usually is lower because of less orders and the inventory holding cost is lower due to
a smaller total EOQ in the centralized than in the decentralized system (cf. appendix C).
Savings in safety stock through inventory pooling stem from balancing demand variabili-
ties (risk pooling).
Although some researchers811 claim the SRL estimated real savings well, in 13 of 14
practical cases we reviewed it overestimated actual inventory savings often significantly.
This means its assumptions are not fulfilled in these cases or there were other inventory- or
service-related changes. In a survey that we conducted with eleven companies from differ-
ent countries and industries (see table D.1) only one fulfilled all, two fulfilled none of the
five assumptions necessary for applying the SRL to regular stock. The participants fulfilled
at least two and at the most seven of the eleven assumptions of the SRL when applied to
safety stock.
It seems that seldom is it appropriate to apply the SRL, because mostly not all its as-
sumptions are fulfilled. Nonetheless, the other SRL- and PE-models that we synoptically
compared in terms of their assumptions in table D.2 might be applied as their assumptions
are fulfilled.
Researchers questioning the SRL either question its assumptions or do make other as-
sumptions and therefore arrive at different results.
Zinn et al.'s (1989) PE shows the reduction in aggregate safety stock by centralizing
several warehouses' inventories in one warehouse in percent, if lateral transshipments are
not usual, there is no lead time uncertainty, customer service level is not affected by a
change in the warehouse number, demand at each warehouse is normally distributed, and
the order quantity equals demand during lead time.
The inventory turnover curve shows average inventory in dependence of the inventory
throughput for a specific company. It can be constructed from a company's stock status
reports and be used to estimate the average inventory (not only safety stock as with the PE)
for any (planned) warehouse throughput without the limitations of the SRL.
(2) Virtual pooling extends a company's warehouse or warehouses beyond its or their
physical inventory to the inventory of other own or other companies' locations by means of
information and communication technologies, drop-shipping, and cross-filling.
(3) Transshipments are inventory transfers among locations (e. g. between warehouses
or stores) for example in case of a stockout.
811
For example Sussams (1986: 10), Pfohl (1994: 141).
128
6 Conclusion
(4) Centralized ordering places joint orders for several locations and later allocates the
orders (perhaps by a depot) to the requisitioners or distribution points in consolidated dis-
tribution according to current demand information.
(5) Order splitting is partitioning a replenishment order into multiple orders with mul-
tiple suppliers or into multiple deliveries.
(6) Component commonality designs products sharing parts or components. This re-
duces the variability of demand for these components.
(7) Postponement delays decisions in production, logistics, or distribution as long as
possible, e. g. developing, purchasing, ordering, fulfillment assignment, production, manu-
facturing, assembly, packaging, labeling, pricing, and shipping. Because of a shorter fore-
cast period and an aggregate forecast more accurate information can be used.
(8) Capacity pooling is consolidating production, service, or inventory capacities of
several facilities. Without pooling every facility fulfills demand just with its own capacity.
With pooling demand is aggregated and fulfilled by a single (perhaps virtually) joint facili-
ty. A higher service level can be attained with the same capacity or the same service level
can be offered with less capacity.
(9) Product pooling is unifying several product designs to a single generic or universal
design or reducing the number of products thereby serving demands that were served by
their own product variant before with fewer products.
In (10) product substitution one tries to make customers buy another alternative prod-
uct, because the original customer wish is out of stock or although it is available (demand
reshape).
Using contingency theory, we explored conditions that favor these individual risk
pooling methods, their advantages and disadvantages, and basic trade-offs in detail. Based
on this synopsis the Risk Pooling Decision Support Tool (RPDST) was developed for
choosing an appropriate risk pooling strategy for a specific situation.
This tool was effectively utilized to choose suitable risk pooling methods for a large
German paper distributor. This application showed that risk pooling can be used effec-
tively as a guiding umbrella principle in systematically restructuring a distribution system
to cope with supplier lead time, customer demand, and forecast uncertainty, as well as
product variety and reduce costs and inventory, increase customer service, and make a
company more competitive. Human resource factors, transaction costs, and local subopti-
mization seem to be important in risk pooling and transshipments. These factors are com-
monly neglected in the risk pooling literature. Risk pooling methods can be and are used
129
6 Conclusion
side by side to take advantage of their benefits, while making up for disadvantages of other
ones. This is in line with our survey findings.
Centralized ordering can reduce average cycle inventory even if a stock-to-demand
order policy is followed and minimum order and saltus quantities have to be observed. It
also enables to decrease safety stock and improve inventory availability as the aggregate
forecast for the central order is more accurate than the disaggregate one for the decentral
orders.
The literature review and this application showed the theoretical benefits risk pooling
can bring. However, survey research on the knowledge and application of risk pooling is
scarce. Therefore, we explored whether the different risk pooling concepts are known, ap-
plied, and associated in a survey of 102 German manufacturing and trading companies:
A lot of respondents do not perceive all the questioned concepts as risk pooling ones. It
appears that mainly transshipments and postponement are associated with risk pooling.
Although at least in our sample the risk pooling concepts are known fairly well (central
ordering, product substitution, and selective stocking the most), only selective stocking,
transshipments, and central ordering are widely applied. Transshipments are more, post-
ponement and product pooling less applied than suggested by some publications mostly for
other regions.
Less sample companies have centralized their warehouse system than European ones.
For our sample we cannot confirm a major decentralization trend as one may expect due to
increasing fuel costs. Companies seem to be consistent in their strategy of centralization,
decentralization, or no change in warehouse number over time.
The utilization and knowledge of a risk pooling concept are correlated. This suggests
that improving the knowledge of risk pooling may increase its application. Research needs
to convey under what conditions and how the different risk pooling concepts can be ap-
plied successfully. This research constitutes one step in this direction.
Prominently associated is past decentralization with future decentralization, the utiliza-
tion of product substitution and demand reshape, past centralization and decentralization
(the only negative correlation), the application of the inventory turnover curve and trans-
shipments, commonality and product pooling, risk pooling and transshipments, turnover
curve and order splitting, selective stocking and virtual pooling, transshipments and virtual
pooling, risk pooling and postponement, product substitution and transshipments, product
substitution and virtual pooling, as well as postponement and commonality and vice versa.
130
6 Conclusion
We can support Rabinovich and Evers' (2003b) finding that time postponement (emer-
gency transshipments and inventory centralization) contributes to the implementation of
form postponement at best only weakly for our sample.
Trading companies seem to apply product substitution and transshipments more than
manufacturing companies. The opposite is true for commonality and postponement. This
was considered in the RPDST. In contrast to Van Hoek (1998b) our sample shows no sig-
nificant difference in the application of postponement by electronics, automotive, food, and
clothing manufacturers.
More large companies appear to use risk pooling and postponement than smaller ones,
perhaps because they can better afford to invest in expensive risk pooling methods as sug-
gested by Huang and Li (2008b: 12). More small companies have centralized their ware-
house or logistics system in the past than large ones in our sample.
The survey is of limited representativeness and generalizability, as a genuine simple
random sample of the relevant population of German manufacturing and trading compa-
nies could not be drawn. Survey research is very laborious and prone to many biases and
errors and has low response rates that disappoint the researcher and impair the results. Our
experience supports that face-to-face methods tend to achieve higher response rates.
As this treatise constitutes a first attempt to structure the vast risk pooling research in
business logistics, develop tools to choose between the various risk pooling methods, and
get an idea about their recognition and application in German companies, we make the
following suggestions for further research:
Our RPDST and its underlying contingency factors should be empirically validated
with additional companies in different industries. Conditions and implementation modali-
ties favoring the different risk pooling methods should be further explored, especially for
product pooling, central ordering, (particularly non-manufacturing) capacity pooling, vir-
tual pooling, and substitution/demand reshape, as our considerations are not exhaustive.
The effectiveness and efficiency of the different methods should be compared. Evers
(1999), Swaminathan (2001), Eynan and Fouque (2005), and Wanke and Saliby (2009)
made a good start here.
Perhaps a normative model could be developed that integrates and compares the ten
identified methods in terms of their costs and benefits in order to choose an optimal one.
However, such a model might be either too complex or simple. Perhaps combining simula-
tion and analytic frameworks may better account for the complicated interaction effects
131
6 Conclusion
among various factors.812 Centralized ordering and transshipments at Papierco could be

simulated as well.
In general, risk pooling models should become more realistic in terms of their assump-
tions, accessibility for practitioners, and allowing (practitioners) to quantify the benefits of
risk pooling.813
We considered choosing suitable risk pooling methods for individual companies for cer-
tain economic conditions and sometimes their relation to their immediate supply chain
members at the most. However, a risk pooling strategy that is beneficial for an individual
supply chain member might not be profitable for others or the whole supply chain.814
Therefore, it should not be implemented by one company unilaterally and further re-
search should focus on the supply chain as a whole: Which conditions render a risk pool-
ing method or a combination of risk pooling methods beneficial for all members of a
supply chain? How do you get independent parties to agree on or join a risk pooling strate-
gy? If some supply chain members are better and other ones worse off after risk pooling,
how do you distribute the gains to encourage risk pooling and make it beneficial for every-
body? The contracting side of risk pooling still leaves room for further research.
Another interesting question to explore would be whether the risk pooling methods
make individual companies and the whole supply chain more resilient to external uncer-
tainties other than demand and lead time uncertainty. Sheffi (2006, 2007) explores this
question for several risk pooling methods and individual companies815. Snyder and Shen
(2006: 42) conclude that “[s]upply chain resilience [to supply disruptions and demand un-
certainty] can be improved significantly without large increases in cost” by centralization
(risk pooling) and decentralization (risk diversification). How should supply chains or lo-
gistics systems be designed, so that risk pooling is exploited optimally and uncertainties
are minimized?
812
Cf. Hwarng et al. (2005).
813
Cf. Heil (2006: 81, 169).
814
Some or all supply chain members may be hurt by centralized ordering (Gürbüz et al. 2007: 305), inven-
tory pooling (Anupindi and Bassok 1999, Pasin et al. 2002, Simchi-Levi et al. 2008: 234-238), postpone-
ment (García-Dastugue and Lambert 2007), transshipments (Grahovac and Chakravarty 2001, Dong and
Rudi 2002, 2004, Zhang 2005, Shao et al. 2009), and virtual pooling (Randall et al. 2002: 56).
815
He examines how an enterprise can be made more resilient to disruptions (supply shortages, production
disruptions, energy price increases, exchange rate fluctuations, strikes, and natural disasters) especially by
risk pooling (Sheffi 2006: 117f., 210, 212, 248, 303, 317), flexibility (Sheffi 2006: 117ff., 199ff.), post-
ponement (Sheffi 2006: 297), commonality (Sheffi 2007: 184-186), standardization (Sheffi 2007: 186-
193), interchangeability (Sheffi 2006: 210), multiple suppliers (Sheffi 2007: 99f., 215-222), invento-
ry/forecast aggregation (Sheffi 2006: 117f.), exchange of parts (Sheffi 2006: 248), and reduction of prod-
uct variants (Sheffi 2006: 119).
132
6 Conclusion
An empirical analysis of risk pooling in the German auto industry might be interesting,
because of the current changes in model mix (high gas prices), production volume (reces-
sion), and steel prices. How can you prepare your production system for something like
this?
How does the current economic downturn affect risk pooling? We already noted that
this might have contributed to increased research and implementation of risk pooling.
How do increased transportation costs affect risk pooling? If they are lasting and
warehouse networks are therefore decentralized, how is this decision affected by risk pool-
ing and how does network redesign affect risk pooling?
Another more representative survey with a higher response rate could also shed light
onto the companies' reasons for and against as well as the manner, degree, benefits, and
costs of applying the risk pooling concepts, as we intended in the six-page version of our
survey that yielded a uselessly low response rate. As risk pooling can affect the whole
supply chain, another survey could consider a supply chain wide perspective rather than
single companies.
Risk pooling is usually shown or proved with the standard deviation. For other meas-
ures of dispersion, such as the variance of independent random variables or the range,
there might be no risk pooling effect. Therefore one could consider risk pooling with
measures of dispersion other than the standard deviation. Risk pooling should also be fur-
ther explored with different order policies.
Finally, the SRL- and PE-models can be extended to include further assumptions. One
should also check whether the conditions or assumptions necessary for the SRL to hold
(necessary conditions) are also sufficient conditions, i. e. if the SRL holds, these conditions
also automatically apply.
133
Appendix A: A Survey on Risk Pooling Knowledge and
Application in 102 German Manufacturing and Trading
Companies
134
1 Introduction
After reviewing the limited recent survey research in OR, business logistics, SCM, and particu-
larly risk pooling mostly with low response rates, we describe our findings from a survey on
risk pooling knowledge and application we conducted among 102 German manufacturing and
trading companies. Such a survey has been demanded by some researchers and – to our know-
ledge – has not been conducted before.
1.1 Motivation: Scarce Survey Research on Risk Pooling
There is little empirical and especially survey research in OR816, business logistics817, and risk
pooling818. The samples and response rates are usually small or not mentioned819.
No surveys have been conducted on risk pooling in business logistics in general yet.
Only few surveys touch types of risk pooling. Most of these surveys deal with postpone-
ment, although the majority of empirical research on postponement relies on case stu-
816
Cachon (2004), Netessine (2005).
817
Griffis et al. (2003: 237).
818
Labro (2004: 358).
819
The following researchers conducted surveys on logistics topics with the following returns: Jackson
(1985, 53 usable responses), Dubelaar et al. (2001, 72 usable of 101 surveys), Griffis et al. (2003: 241f.,
response rate: 13.25 %), Kisperska-Moroñ (2003: 130, 538 questionnaires, on average three respondents
in one company), Bagchi and Skjoett-Larsen (2005, 149 responses, response rate: 5.7 %), White (2005,
429 answers, response rate: 10.93 %), Hilmola and Szekely (2006, 67 usable responses, response rate:
8.9 %). Johnson et al. (2006, 297 usable responses, response rate: 55 %, 305 usable responses, response
rate: 51 %, 284 usable responses, response rate: 44 %), Shao and Ji (2006, 321 surveys, response rate:
89.2 %), Böhnlein and Lünemann (2008, 51 responses), Chikán (2009, 51 responses, response rate:
20.82 %).
Surveys on postponement received 65 (Kisperska-Moroñ 2003: 130), 76 (Yang et al. 2005b), 80 (Van
Hoek 1998b), 102 (Chiou et al. 2002), 106 (Huang and Li 2008b), 111 (completed 24-page workbooks in
interviews with world class companies, CLM 1995: 8, 384), 129 (McKinnon and Forster 2000: 5, Delphi
survey with logistics experts), 256 (Rabinovich and Evers 2003b), 306 (Nair 2005), 322 (in the base-line
survey in Germany, CLM 1995: 381f.), 358 (Oracle 2004, response rate: 2.24 %), 782 (Van Hoek 2000b,
Van Hoek and Van Dierdonck 2000, response rate: 53 %), and 3,693 (altogether in the base-line survey,
CLM 1995: 381f.) usable responses. The surveys with high returns were conducted via logistics associa-
tions (CLM 1995, Rabinovich and Evers 2003b, Oracle 2004, Nair 2005) or via phone with very few
questions (Van Hoek 2000b). They were administered in Europe (McKinnon and Forster 2000: 5),
Greater China (Mainland China, Hong Kong, and Taiwan) (Huang and Li 2008b), Taiwan (Chiou et al.
2002), the U. S. (an expert study by Morehouse and Bowersox 1995, Van Hoek 1998b, Rabinovich and
Evers 2003b, Nair 2005), the UK (Yang et al. 2005b), the Netherlands (Van Hoek 2000b, Van Hoek and
Van Dierdonck 2000), and Upper Silesia, Southern Poland (Kisperska-Moroñ 2003) among manufactur-
ing (Van Hoek 1998b, Van Hoek and Van Dierdonck 2000, Chiou et al. 2002, Kisperska-Moroñ 2003,
Rabinovich and Evers 2003b, Nair 2005, Yang et al. 2005b, Huang and Li 2008b), trading (Van Hoek and
Van Dierdonck 2000, Nair 2005), and transport and logistics services companies (Van Hoek 2000b, Van
Hoek and Van Dierdonck 2000). Graman and Magazine (2006: 1068, 1070f., 1075) conducted depth in-
terviews with five managers of a small U.S. manufacturer of mops and brooms on issues influencing suc-
cessful postponement implementation.
135
Appendix A
dies820, such as Dapiran (1992), Cooper (1993, examples), Feitzinger and Lee (1997), Garg
and Tang (1997), Van Hoek (1997, 1998a, 1998b), Van Hoek et al. (1998), Magretta
(1998), Brown et al. (2000), Waller et al. (2000), Ghemawat and Nueno (2003), Huang and
Lo (2003), and Huang and Li (2008a).
Whereas some researchers find postponement application has increased, will increase,
or postponement is widely applied821, others state it is not applied as much (as expected)822
and will be even less used in the future823. This is explained by “a general lack of under-
standing of postponement” regarding its conception, gains, and costs824, perceived technol-
ogy limitations, ineffective organization alignment825, as well as difficulties in managing
the relations to suppliers and customers826 and in estimating the costs and benefits of post-
ponement827.
Van Hoek (1998b: 513) finds that “[d]ownstream activities such as distribution and
packaging are largely postponed (56.93 per cent of the flow of goods and 53.95 per cent).
Upstream activities such as engineering and purchasing are postponed to a lesser extent
(37.49 per cent of the flow of goods and 37.42 per cent). Overall however, the findings
indicate that postponement is applied at multiple points in the chain and to a significantly
high share of the flow of goods”. Yang et al. (2005b: 1000) remark that postponement is
mostly applied as make to order and ship to order.
820
Van Hoek (2001).
821
CLM (1995: 213), Morehouse and Bowersox (1995), Van Hoek (1998b, 2000b), Van Hoek and Van
Dierdonck (2000), McKinnon and Forster (2000: 7f.), Chiou et al. (2002), Huang and Li (2008b). CLM
(1995: 213) finds “Research clearly supports the generalization that firms are rapidly adopting both form
and time postponement”, although 10.3 % of North American and 85 % of non-North American world
class companies have decreased or substantially decreased the use of postponement and 31 % of North
American and 0 % of non-North American (European/Pacific Basin) world class companies have in-
creased or substantially increased the use of postponement (Workbook Section 4: Organization, Question
31, “Electronic Appendix” Disk to CLM 1995). Yet, only few non-North American companies completed
the workbook (CLM 1995: 384).
822
Van Hoek et al. (1998: 34, 51), Bowersox et al. (1999), Battezzati and Magnani (2000: 423), Brown et al.
(2000: 78), Van Hoek (2000b, in third party logistics or TPL), Van Hoek and Van Dierdonck (2000, in
TPL), Oracle (2004), Yang et al. (2004, 2005a, 2005b: 991), Boone et al. (2007: 594), Yang et al. (2007:
972). Van Hoek et al. (1998: 51, 34) observe that although “the theoretical benefits of postponement
[were] introduced [already] in the 1950s and 60s”, this “highly attractive principle has hardly been ap-
plied to date”. “[P]ostponement applications are still in the infancy stage” (Van Hoek et al. 1998: 51). It
“has an enormous potential, still largely under-exploited, in Italy” (Battezzati and Magnani 2000: 423).
Boone et al. (2007: 594) remark “the slow rate of postponement adoption among practitioners”.
„[P]ostponement is an underutilized concept“ (Yang et al. 2007: 972).
823
“Postponement applications are still not as widespread as expected. […] The respondents also expected
postponement to be less used in three years” (Yang et al. 2005b: 991).
824
Oracle (2004: 2, 7).
825
Capgemini (2003), Matthews and Syed (2004: 30).
826
Yang et al. (2005b: 991), Waller et al. (2000), Van Hoek et al. (1998).
827
Van Hoek et al. (1998).
136
Appendix A
Rabinovich and Evers (2003b: 41f.) find that time postponement (emergency trans-
shipments) contributes to the implementation of form postponement.
Huang and Li (2008b: 5, 16f.) discover that the degree of postponement application is
higher in China than CLM (1995), Yang et al. (2005a), and Van Hoek (1998b) remarked in
Western countries and increases. Chiou et al. (2002: 122) also conclude “that form post-
ponement strategies are practiced widely by IT firms in Taiwan”. Huang and Li (2008b:
12) sent their questionnaires only to 411 companies that had confirmed on the phone that
they applied postponement. They, Chiou et al. (2002), and Yang et al. (2005b) do not state
how many companies in their research countries actually apply postponement. In a person-
al correspondence on September 4, 2009 Dr. Biao Yang, The York Management School,
University of York, UK, explained that speaking from their experience, given the working
definitions of different postponement strategies they provided828, very few companies
would claim that they are not using any postponement.
The Council of Logistics Management (CLM) conducted a base-line survey in North
America (Canada and the U.S.), Europe (France, Germany, the Netherlands, Norway,
Sweden, and the UK), and the Pacific Basin (Australia, Japan, and Korea) from May to
September 1993. Of the 21,592 mainly manufacturing companies contacted, 3,693 an-
swered (response rate: 17.1 %). In Germany 3,000 surveys were mailed and 322 returned
(response rate: 10.7 %).829
It also did interviews with 111 companies from 17 countries in North America, Europe,
and the Pacific Basin from early 1994 until spring 1995 that were believed “by a group of
logistics experts to have a high potential to possess world class logistical capabilities” and
asked them to complete a 24-page workbook.830
35.3 % of North American and 29.1 % of non-North American world class companies
consider location flexibility (“[t]he ability to service customers from alternative ware-
house locations”, i. e. in our opinion transshipments or virtual pooling) less or least impor-
828
In their survey Yang et al. (2005b: 1002f.) ask for “the percentages of goods related to” engineer, pur-
chase, make, final manufacturing/assembling, package/label, and ship to order to elicit the extent of post-
ponement application in the respondents' companies. Thus they equate make-to-order with postponement
or delayed differentiation, while Cachon and Terwiesch (2009: 344) distinguish these strategies: Al-
though “they are conceptually quite similar”, “[d]elayed differentiation is thought of as a strategy that
stores nearly finished product and completes the remaining few production steps with essentially no de-
lay. Make-to-order is generally thought to apply to a situation in which the remaining production steps
from components to a finished unit are more substantial, therefore involving more than a trivial delay.
Hence, delayed differentiation and make-to-order occupy two ends of the same spectrum with no clear
boundary between them”.
829
CLM (1995: 381f.). CLM was named Council of Supply Chain Management Professionals (CSCMP) in
2005.
830
CLM (1995: 8f., 384).
137
Appendix A
tant, 35.3 % or 37.5 % respectively more or most important. The North American and non-
North American world class companies see mostly the logistics department as responsible
for location flexibility (77.19 and 74.76 % responsibility). Fifty-one percent of North
American and 58.3 % of non-North American world-class companies believe they perform
better or much better than their competitors in terms of location flexibility.831
According to McKinnon and Forster (2000: 17), “the proportion of retail sales bypass-
ing conventional shops and being delivered directly to the home is likely to increase signif-
icantly” until 2005. “Direct distribution to the home from nationally based suppliers is like-
ly to experience the fastest growth”832. That means that virtual pooling in the form of
drop-shipping is expected to increase.
In a small phone survey by Randall et al. (2006) of 64 (56 responses, 54 usable re-
sponses, response rate: 84.4 %) publicly held e-tailers that made up 60 % to 70 % of U.S.-
wide e-tailing revenue, 36 companies (67.7 %) chose to hold inventory. That means 33.3 %
of the surveyed e-tailers conducted virtual pooling.
In the CLM base-line survey, 62.81 % of North American, 68.31 % of European, and
47.26 % of Pacific Rim companies agree or strongly agree that on an equal volume basis
they had inventory located at fewer sites in 1993 (today) than in 1988 (five years ago).833
That means that 68.31 % of the surveyed European companies conducted inventory pool-
ing.
The world class companies' “[l]ogistics strategy includes a priority to reduce […] [t]he
number of logistics facilities” more so in Europe and the Pacific Basin, where the mean
answer was 3.55 than in North America, where the mean answer was 3.23 on a scale from
1 (Strongly Disagree) to 5 (Strongly Agree).834
McKinnon and Forster (2000: 6f.) find “The centralisation of inventory is likely to out-
strip that of production. At a European level, the degree of inventory concentration is fore-
cast to increase by a third, twice as much as at a national level. A significant minority of
the panel (around 15 %) indicated that within countries there would be net decentralisation
of inventory, with firms increasing their number of stockholding points. The prevailing
view, however, was that inventory centralisation, which has been one of the main logistical
trends over the past 30 years, would continue apace for at least the next 5 years”.
831
Workbook Section 3: Relative Performance Competencies, Question 20, “Electronic Appendix” Disk to
CLM (1995).
832
McKinnon and Forster (2000: 9).
833
Logistics Practices Results, Question 7, “Electronic Appendix” Disk to CLM (1995).
834
CLM (1995: 387).
138
Appendix A
A Warehousing Education and Research Council survey with 139 U.S. participants in
manufacturing, retail, wholesale, public utilities and government showed that the number
of warehouses continues to shrink, although there are a few new large facilities in West or
Mid Atlantic regions.835
Contrarily, according to a survey by Lemoine and Skjoett-Larsen (2004) among logis-
tics managers of mechanical, electronic, and medical equipment companies in Denmark in
2001 (46 usable questionnaires, response rate: 10 %) the number of suppliers, production
facilities and warehouses has not been changed in the last three to five years in Europe, but
has been increased or will be increased outside Europe.
Bandy (2005) conducts an anonymous survey among 24 students that shows that a sim-
ple spreadsheet in-class simulation helps students understand the impact of pooling safety
stock.
Reviewing literature and interviewing Belgian companies, Naesens et al. (2007) find
that companies are reluctant to implement inventory pooling in a horizontal collaboration.
Nonetheless, companies seem to have conducted inventory pooling within companies to
a broad extent.
Twenty-seven percent of North American and 29.1 % of non-North American world-
class companies consider substitution flexibility (“[t]he ability to substitute product or
service offerings in the event of a delay or stockout versus backorder or line cancellation”)
less or least important, 46.2 % or 25 % respectively more or most important. Responsibili-
ty for substitution flexibility rests 49.9 % with logistics, 35.63 % with marketing according
to North American, or 48.41 % or 42.95 % respectively according to non-North American
world-class companies. 44.2 % of North American and 37.5 % of non-North American
world-class companies believe they perform better or much better than their competitors in
terms of substitution flexibility.836
Only few surveys provide information on how many companies apply the respective
risk pooling method.
Nearly 50 % of more than 350 companies from all over the world have not implemented
postponement strategies.837 Hence, about half of these companies use postponement.
More than 40 % of the companies surveyed by Kisperska-Moroñ (2003: 131) gear total
production towards customer orders. Surveyed companies manufacture on average 16 % to
835
Modern Materials Handling (2002).
836
CLM (1995).
837
Oracle (2004).
139
Appendix A
stock and 84 % to order, engineering 97 %, metal 74 %, plastic products 83 %, and electro-

technical companies 77 % to order. Thus one could infer that about 40 % of the surveyed
companies in Upper Silesia, Poland apply manufacturing postponement.
From 1988 to 1993 inventory was centralized or pooled by 68.31 % of the European
companies surveyed by the CLM.838
Virtual pooling is applied by 33.3 % of the U.S. e-tailers that Randall et al. (2006) in-
terviewed.
None of the surveys gives detailed information on Germany, although Germany is the
fourth largest economy in the world according to its nominal GDP in USD in 2008839 and
is very progressive, advanced, and sophisticated in logistics, especially in “operating effi-
ciency and cost” and customer service, and, like other European countries, in the “func-
tional integration of logistics work within the firm” and “environmental issues” like “re-
verse logistics”840. European countries have opportunities for improvement in “the external
integration of logistics systems and development of supply chain relationships throughout
Europe”841.
1.2 Objective
We would like to determine to what extent the different risk pooling concepts are known and
applied by 102 German manufacturing and trading companies and examine for our sample the
following remarks, suggestions, and demands:
Van Hoek (1998b) finds that postponement is extensively applied in electronics and au-
tomotive and less extensively applied in food and clothing. We would like to find out
whether this applies to our sample as well.
Alfaro and Corbett (2003: 12f., 15) observe that “several approaches to the widely rec-
ognized challenge of managing product variety rely on the pooling effect. Pooling can be
accomplished through the reduction of the number of products or stock-keeping units
(SKUs), through postponement of differentiation, or in other ways [component commo-
nality and inventory pooling]. These approaches are well known and becoming widely
applied in practice”. However, there are no studies yet whether product pooling and com-
ponent commonality really are widely applied in practice.
838
839
IMF (2009).
840
CLM (1995: 365).
841
CLM (1995: 366).
140
Appendix A
Herer et al. (2006) remark “Transshipments are not widely adopted in practice because
the IT infrastructure is inadequate and there are no realistic models to take advantage of
transshipment benefits”. However, there are no studies yet whether transshipments really
are not widely used in practice.
Thomas and Tyworth (2006) demand a survey to determine whether order splitting is
employed in practice842.
Boone et al. (2007) review postponement literature published from 1999 to 2006 and
conclude research should be extended to non-manufacturing postponement and investigate
the slow rate of postponement adoption among practitioners.
Huang and Li (2008b: 12) only surveyed large electronic/information technology, cloth-
ing, and electric appliances companies. They remark large companies may apply post-
ponement more than small ones, since they may be better able to afford the redesign of
products and/or processes usually required by postponement843. “Hence, further studies
could compare the differences of postponement application between large-scale and small
and medium companies”844.
842
“Yet, despite the impressive amount of past and present work on pooling lead-time risk by order splitting,
it is difficult to find any substantive evidence of successful applications in academic or practitioner litera-
ture” (Thomas and Tyworth 2006: 246). “Although studies in both tracks provide hypothetical evidence
that pooling lead-time risk by splitting orders simultaneously may be worthwhile, it is difficult, if not im-
possible, to find substantive empirical validation of the proposition” (Thomas and Tyworth 2006: 254).
“Collectively, these limitations make the value proposition rather dubious. Thus future research in this
area should include case studies, simulations, and surveys to determine whether companies use such order
splitting methods and the business setting where successful applications appear” (Thomas and Tyworth
2006: 254f.).
843
Ernst and Kamrad (2000), Aviv and Federgruen (2001b).
844
Huang and Li (2008b: 19).
141
2 The Survey
There has not been a survey on how many companies know and apply the various risk pooling
methods yet, especially not in Germany. In order to respond to the aforementioned demands
and remarks, we conducted a survey among German manufacturing and trading companies
from September 2008 to August 2009 and obtained 102 completed questionnaires (response
rate: 21.3 %) from logistics or supply chain management professionals. Since an earlier six-
page survey that was e-mailed to 3,600 companies had a response rate of only 0.3 %, we asked
478 manufacturing and trading company representatives in person to complete the one-page
questionnaire in figure A.4 or forward it to the appropriate person mainly on company and job
fairs in various German cities.
Respondents were promised a summary of the survey results as an incentive.845 Our ex-
perience supports that face-to-face methods usually achieve higher response rates846.
2.1 Research Design
The respondents worked for companies from the following branches of economic activity ac-
cording to the German Classification of Economic Activities Edition 2003 of the German Fed-
eral Statistical Office (Statistisches Bundesamt Deutschland):
845
Cf. Yang et al. (2005b: 996).
846
Kisperska-Moroñ (2003: 130), Groves et al. (2004: 165).
142
Appendix A
The whole Responding

a
population companies
Classification Description Number % Number % Variance (%)

Extraction of crude petroleum and natural gas;
11 service activities incidental to oil and gas extraction, 17 0.01 1 0.98 0.97
excluding surveying
15 Manufacture of food products and beverages 5,189 4.46 4 3.92 -0.54
16 Manufacture of tobacco products 24 0.02 1 0.98 0.96

Manufacture of wearing apparel; dressing and
18 378 0.32 1 0.98 0.66
dyeing of fur
21 Manufacture of pulp, paper and paper products 817 0.70 2 1.96 1.26
Publishing, printing and reproduction of recorded
22 2,456 2.11 2 1.96 -0.15
media
24 Manufacture of chemicals and chemical products 1,405 1.21 3 2.94 1.73
25 Manufacture of rubber and plastic products 2,635 2.26 2 1.96 -0.30
26 Manufacture of other non-metallic mineral products 1,677 1.44 1 0.98 -0.46
27 Manufacture of basic metals 897 0.77 1 0.98 0.21

Manufacture of fabricated metal products, except
28 6,175 5.31 4 3.92 -1.38
machinery and equipment
29 Manufacture of machinery and equipment n.e.c. 5,990 5.15 13 12.75 7.60
30 Manufacture of office machinery and computers 164 0.14 1 0.98 0.84

Manufacture of electrical machinery and apparatus
31 1,922 1.65 7 6.86 5.21
n.e.c.
Manufacture of radio, television and communication
32 542 0.47 5 4.90 4.44
equipment and apparatus
Manufacture of medical, precision and optical
33 2,067 1.78 2 1.96 0.18
instruments, watches and clocks
Manufacture of motor vehicles, trailers and semi-
34 1,001 0.86 14 13.73 12.87
trailers
35 Manufacture of other transport equipment 314 0.27 3 2.94 2.67
36 Manufacture of furniture; manufacturing n.e.c. 1,484 1.27 2 1.96 0.69
37 Recycling 168 0.14 3 2.94 2.80
40 Electricity, gas, steam and hot water supply 1,196 1.03 2 1.96 0.93
Building of complete constructions or parts thereof;
45.2 6,352 5.46 1 0.98 -4.48
civil engineering
51.46 Wholesale of pharmaceutical goods 3,148 2.70 2 1.96 -0.74
Wholesale of paper and paperboard, stationery,

51.47.8 2,259 1.94 3 2.94 1.00
books, newspapers, journals and periodicals
Non-specialized wholesale of wood, construction
51.53.1 1,254 1.08 2 1.96 0.88
materials and sanitary equipment
Wholesale of agricultural machinery and
51.88 1,333 1.15 1 0.98 -0.16
accessories and implements, including tractors
Retail sale in non-specialized stores with food,
52.11 24,790 21.30 7 6.86 -14.44
beverages or tobacco predominating
52.42 Retail sale of clothing 25,896 22.25 10 9.80 -12.44
52.46 Retail sale of hardware, paints and glass 11,279 9.69 1 0.98 -8.71
52.48.6 Retail sale of games and toys 3,564 3.06 1 0.98 -2.08
Total 116,393 100 102 100
Table A.1: Comparing Respondents with the Total Population of Manufacturing and Trading
Companies in Germany
a
Source: Statistisches Bundesamt Deutschland (2009i: 371, 408f.). Companies with in general 20
employees and more. Status of 09/30/2006 (trade), 09/30/2007 (manufacturing).
143
Appendix A
As 78.7 % of the sampling frame population did not respond and the sample is small, it is not
too representative of neither the whole population in the different branches of economic ac-
tivity the respondents' companies belong to in table A.1 nor of the target population of German
manufacturing and trading companies.847 Not all areas of domestic production and trade are
covered. Our sample consists of too many manufacturing companies, especially from vehicle
construction (34, 35), although this is an important industry in Germany, manufacture of ma-
chinery (29, 31, 32), and recycling (37) and too few trading companies, particularly in food
(52.11), clothing (52.42), and hardware retail (52.46) as the variance in the last column shows.
Deviations of responding companies from the whole population in Huang and Li
(2008b: 13) are of similar magnitude. In contrast, the responding companies in Yang et al.
(2005b: 997) closely reflect the whole population. Of the 28 mentioned surveys in this sec-
tion848 only Huang and Li (2008b) and Yang et al. (2005b) (7.14 %) compare the respond-
ing companies' characteristics to the whole population (retrospective quota sampling).
We have to highlight that the whole population in table A.1 only consists of companies
with 20 or more employees, while our sample encompasses three companies (respondents
2, 65, and 100 from industrial sectors 52.42, 27, and 51.47.8) with less than 20 employees.
If we omitted them, the variance would increase and the representativeness decrease even
more. A detailed classification of the branches of economic activity with all German com-
panies with any number of employees is not publicly available. The German Federal Statis-
tical Office told us we could apply for a special query with charges. A retrospective quota
sampling with more detailed subgroups of industrial economic activity than in table A.1
and all German companies, which cannot be conducted due to lacking data, might reveal
that our sample is more representative of these subgroups.
The size of the responding companies ranges from 5 to 182,739 employees. 2 % have
less than 10, 9 % less than 50, 31 % less than 250, 32 % less than 500, and 68 % have 500
847
There were 52,362 manufacturing (9.94 % of manufacturing and trading companies) and 474,065
(90.06 %) trading companies with in general 20 or more employees in Germany in 2007 or 2006 respec-
tively (Statistisches Bundesamt Deutschland 2009i: 409, 371). In our sample there are 75 (73.53 %) man-
ufacturers and 27 (26.47 %) trading companies.
848
Jackson (1985), CLM (1995), Morehouse and Bowersox (1995), Van Hoek (1998b, 2000b), McKinnon
and Forster (2000), Van Hoek and Van Dierdonck (2000), Dubelaar (2001), Chiou et al. (2002), Griffis et
al. (2003), Kisperska-Moroñ (2003, two surveys), Rabinovich and Evers (2003b), Lemoine and Skjoett-
Larsen (2004), Oracle (2004), Bagchi and Skjoett-Larsen (2005), Nair (2005), White (2005), Yang et al.
(2005a, 2005b), Himola and Szekely (2006), Johnson (2006), Randall et al. (2006), Shao and Ji (2006),
Naesens et al. (2007), Böhnlein and Lünemann (2008), Huang and Li (2008b), and Chikán (2009).
144
Appendix A
or more employees. The majority of the sample consists of larger companies, while the
majority of the whole population comprises small ones.849
Nevertheless our survey can at least show the risk pooling knowledge and activities of
the surveyed companies for a given moment in time. They are relatively heterogeneous in
terms of industry membership and number of employees (size), so that risk pooling know-
ledge and activities of a wide company spectrum and size specific specialties can be de-
tected.850 Besides, the respondents do not deviate from the overall population extremely
regarding the branches of activity. Therefore, the absence of data from the nonrespon-
dents may not contaminate the conclusions drawn from the sample too much.851
The questionnaire was scrutinized by three academics and three logistics managers,
pretested among eleven companies, and adjusted according to the suggestions and de-
tected problems before the final distribution to insure content validity.852 Validity meas-
ures whether the questions capture what they are intended to capture.853 Of the 28 surveys
mentioned previously 28.57 % deal with their validity. Reliability measures “whether res-
pondents are consistent or stable in their answers”854. Only 17.86 % of the 28 surveys men-
tioned address their reliability. We did not measure reliability as respondents usually are
not willing to be interviewed twice or to answer a questionnaire enlarged by asking for the
same construct twice.855 We could not check for “systematic reporting errors”856 (response
bias), as external data, which the survey responses could be compared to, are not availa-
ble857. None of the 28 surveys mentioned refers to their response bias.
We did not check for non-response bias, because we could not collect the nonrespon-
dents' characteristics858. Non-response bias is the distortion of the sample and therefore the
basic population because some people do not respond or do not answer certain questions.
The answers of respondents may differ from the ones of nonrespondents. There is no con-
sensus on when nonresponse reduces survey quality and therefore on appropriate response
849
In 2006 there were 4,307 (0.63 %) active industrial and 1,362 (0.18 %) trading companies with 250 and
more, 680,561 (99.37 %) or 735,820 (99.81 %) ones respectively with less than 250 employees subject to
social security (Statistisches Bundesamt Deutschland 2009i: 493). In our sample 74.67 % of the industrial
and 51.85 % of the trading companies have 250 and more employees.
850
Cf. Böhnlein and Lünemann (2008: 6).
851
Cf. Vockell and Asher (1995: ch. 8).
852
Cf. Yang et al. (2005b: 995, 998), Huang and Li (2008b: 13).
853
Groves et al. (2004: 254).
854
Groves et al. (2004: 261).
855
Groves et al. (2004: 266), Yang et al. (2005b: 998).
856
Groves et al. (2004: 259).
857
Cf. Groves et al. (2004: 266).
858
Cf. Groves et al. (2004: 169-196), Schneekloth and Leven (2003).
145
Appendix A
rates.859 Nevertheless, our response rate is above the unfavorable response rates below
20 %860, within the 10 % to 30 % of the past mean861 and the 4 % to 32.7 % of large sam-
ples in the Journal of Business Logistics from 1997 to 2001862, and higher than a lot of the
mentioned surveys' ones863.
Nonresponse rates often are ignored, because it is assumed that the reason for the non-
response is not associated with the measured statistical values and therefore it does not
affect the quality of the survey's results.864 Of the 28 surveys we mentioned ten (35.71 %)
deal with their non-response bias. Most of these determine the nonresponse bias, checking
whether the answers of first respondents and late respondents differ, although late respon-
dents may not be an appropriate estimate for nonrespondents.
Nonetheless, the answers of our 51 first and 51 last respondents are independent (the
null hypothesis that there is no significant relationship between the answers cannot be re-
jected) in Pearson's and Yates' chi-square and Fisher's exact test at the 5 % level, except for
the question about their risk pooling knowledge865 and centralization in the past866. In all
other cases the distribution of the answers within the two groups does not differ signifi-
cantly: The two-sided asymptotic significance of the respective test statistic is greater than
0.05 in each case, so that it is safe to say that any differences are due to chance variation,
which implies that the two groups answered equally.
With our sample size of 102 the confidence interval or margin of error is 9.7 on the
95 % confidence level for an estimated percentage of 50 %, i. e. one can be 95 % certain
that if 50 % of the sample picks an answer between 40.3 % (50-9.7) and 59.7 % (50+9.7)
of the whole population would have also picked that answer. However, the true percentage
of the population only is the sample's percentage plus/minus the confidence interval, if a
genuine simple random sample of the relevant population was drawn. This cannot be con-
firmed for our sample, so that this confidence interval is an underestimate. No errors and
“biases of coverage, sampling, nonresponse, or measurement […] are included in the mar-
gin of error”867. When random sampling is not used, one has to depend on logic apart from
859
Fowler (1993), Schneekloth and Leven (2003), Groves et al. (2004: 169-196).
860
Yu and Cooper (1983).
861
Flynn et al. (1990).
862
863
Cf. Yang et al. (2005b: 998).
864
Groves et al. (2004: 169-196), Schneekloth and Leven (2003).
865
In this case Pearson's chi-square test gives χ2 = 9.046 with df = 1 and p = 0.00263, Yates' correction for
continuity χY = 7.880 with df = 1 and p = 0.004998, and Fisher's exact test p (two-tailed) = 0.004728.
866
In this case χ2 = 4.752 with df = 1 and p = 0.029264, χY = 3.928 with df = 1 and p = 0.047432, and
p = 0.046973 in Fisher's exact test.
867
Groves et al (2004: 382).
146
Appendix A
mathematical probability to assess to what extent the sample deviates from the population
it was drawn from.868
2.2 Data Analysis and Findings
2.2.1 Risk Pooling Knowledge and Utilization in the German Sample

Companies
100 91
88
Percent of Responding Companies
90 84
79 79
76
80
70
70 66 66 64 64 66
58
60 54 54 55
49 47 48
50 45 46
40 41
40 34 35 35
28 28
30 25
22
16 17
20
6 7
10
0
Knowledge Utilization
Figure A.1: Risk Pooling Knowledge and Utilization in the German Sample Companies
The most known concepts related to risk pooling in the whole sample are central ordering,
product substitution, selective stocking, transshipments, and order splitting, followed by prod-
uct pooling, commonality, virtual pooling, capacity pooling, “risk pooling”, postponement,
demand reshape, the portfolio effect (PE), inventory turnover curve, and square root law (SRL)
in order of descending percentage of respondents that know these concepts.
Surprisingly, the SRL (28 %) is less known than the PE (54 %), although it is the sim-
pler concept and there are more and earlier publications on it in business logistics. Perhaps
the PE is more popular, as it is a generalization of the SRL and originated in finance.
868
Vockell and Asher (1995: ch. 8).
147
Appendix A
A lot of respondents obviously did not realize that risk pooling may comprise all these
different concepts. Otherwise they would have noted that they know and apply risk pool-
ing, whenever they checked to know or apply any of these concepts. Risk pooling is a
hypernym and not a risk pooling concept itself.
The most applied risk pooling concepts in the whole sample are selective stocking,
transshipments, central ordering, commonality869, and centralization, followed by product
substitution, virtual pooling, capacity pooling, product pooling, postponement, order split-
ting, “risk pooling”, the PE, inventory turnover curve, demand reshape, decentralization,
and the SRL in order of descending percentage of surveyed companies applying these con-
cepts.
While the knowledge of the different risk pooling concepts is fairly good870 (only the
SRL and the inventory turnover curve are known by less than 50 %), they are not widely
applied. Only the well established concepts selective stocking, transshipments, and central
ordering show application rates over 50 %, namely 66 %. This disagrees with Herer et al.
(2006), who remark “Transshipments are not widely adopted in practice because the IT
infrastructure is inadequate and there are no realistic models to take advantage of trans-
shipment benefits”. Some companies which do not employ central ordering remarked it
increased transportation and delivery costs too much. This could be counteracted with a
centralized ordering system like the one we designed for Papierco earlier. Transportation
costs do not increase (substantially) or the supplier incurs them.
Postponement is only applied by 35 % of the companies. Thus we cannot conclude
“that firms are rapidly adopting both form and time postponement”871 and widely apply
it872 in Germany as CLM (1995: 209) reports for North America and Europe, McKinnon
and Forster (2000: 5) for Europe, and Chiou et al. (2002) and Huang and Li (2008b) for
China. More sample companies in Germany (65 %) have not implemented postponement
strategies than suggested by Oracle (2004) (nearly half of 358 mainly manufacturing com-
panies from all over the world). We agree with Van Hoek et al. (1998), Bowersox et al.
(1999), Battezzati and Magnani (2000), Brown et al. (2000: 78), Oracle (2004), Yang et al.
869
Wazed et al. (2009: 70) find “Although the benefits of commonality are widely known, many companies
are still not taking full advantage of it when developing new products or re-designing the existing ones”.
We can agree with them for our sample.
870
Alfaro and Corbett (2003: 12f., 15) already remark that risk pooling approaches such as product pooling,
postponement, component commonality, and inventory pooling “are well known […] in practice”.
871
CLM (1995: 213).
872
148
Appendix A
(2004, 2005a, 2005b, 2007), and Boone et al. (2007) that postponement is not applied as
much (as expected).
Some respondents told us they did not use postponement as it entailed completely dif-
ferent production and storage concepts and therefore usually was uneconomical. Research
has to determine more clearly the changes necessary for postponement, methods for calcu-
lating the costs of postponement, and conditions that render postponement economical.
Practice seems to lack the knowledge to implement postponement economically. Perhaps
more concepts have to be developed that may make postponement less expensive, such as
operations reversal. This is in line with Oracle (2004: 7f.). Our survey suggests that the
reason for the low application may not lie in the basic knowledge of this concept (55 %
know it), but it is less known than most of the other concepts except for demand reshape,
the PE, inventory turnover curve, and SRL.
Component commonality is well known and applied by nearly every second surveyed
company. Thus, we can support Kim and Chhajed (2001: 219), who remark “The use of
commonality in product line extensions is a growing practice in many industries”, and Al-
faro and Corbett (2003: 12, 15), who observe it is “becoming widely applied in practice”.
Nevertheless some companies who do not use it pointed out it was too expensive. This
confirms that the cost of common components can be high873.
Location flexibility achieved by transshipments and virtual pooling seems to be more
important in the sample companies in Germany than suggested by CLM874 for North
American and non-North American world class companies. McKinnon and Forster (2000:
9, 17) predicted virtual pooling in the form of drop-shipping to increase. Randall et al.
(2006) found that 33.3 % of the 56 interviewed U.S. e-tailers relied on virtual pooling.
More generally we find that 46 % of the surveyed manufacturers and traders apply virtual
pooling in Germany. Our sample only contains a single toy e-tailer.
Fewer German sample companies (48 %) have centralized their stockholding than Eu-
ropean ones overall (68.31 %)875. Our findings agree with McKinnon and Forster (2000:
6f.), Modern Materials Handling (2002), and Alfaro and Corbett (2003: 12f.) that inventory
centralization is an important trend. However, we can also confirm the minority opinion
of 15 % of the panelists in McKinnon and Forster (2000: 6f.) that there will be a slight net
decentralization of inventory in Germany in the future. Thirty-five percent of the German
873
874
CLM (1995).
875
149
Appendix A
sample companies have not changed the number of warehouses or production facilities in
the past. Lemoine and Skjoett-Larsen (2004) reached a similar result in Denmark. Some of
the companies that did not or will not centralize declared it increased transportation costs
too much.
Most companies have already centralized or to a lesser extent decentralized their ware-
house or logistics system in the past. Only six percent intend to centralize their logistics
system in the future. Merely seven percent plan to increase the number of warehouse or
production locations in the future. This is in opposition to the expectation that the increas-
ing fuel and transportation costs induce companies to decentralize their warehouse or logis-
tics systems. The risk pooling advantages derived from physical inventory pooling may
outweigh the increased transportation costs.
Product substitution flexibility seems to be more important in Germany than suggested
by CLM876 for North American and non-North American world class companies. Compa-
nies applying demand reshape mostly indicated they persuaded customers to buy another
substitute product even though the original customer wish was available to increase the
profit margin and not for risk pooling reasons as intended by Eynan and Fouque (2003,
2005).
We have to disagree with Alfaro and Corbett (2003: 12) who observe “the reduction of
the number of products or stock-keeping units (SKUs)”, also known as product pool-
ing877, and “postponement of differentiation” are “becoming widely applied in practice”.
Product pooling is applied by 40 % and postponement by 35 % of the companies in our
sample.
We can answer the call of Thomas and Tyworth (2006: 254f.) for empirical research “to
determine whether companies use such order splitting methods”, pointing out that only
35 % of German companies in our sample do apply them. However, our survey does not
inform about the conditions and success of these order splitting applications.
Some interviewees said they did not know when and how it was favorable to apply the
various risk pooling concepts. Research needs to address these issues further. More re-
search is needed to convey to practice when and how the different risk pooling concepts
may be implemented successfully. Our research constitutes one step in this direction.
876
CLM (1995)
877
150
Appendix A
2.2.2 Association between the Knowledge of Different Risk Pooling

Concepts
The knowledge of risk pooling is significantly associated with the knowledge of postponement
(Goodman and Kruskal's Tau (τ) = 0.147, approximative significance based on chi-square ap-
proximation (p) = 0.000), product pooling (τ = 0.104, p = 0.001), transshipments (τ = 0.091,
p = 0.002), the turnover curve (τ = 0.087, p = 0.003), commonality (τ = 0.066, p = 0.010), de-
mand reshape (τ = 0.061, p = 0.013), and virtual pooling (τ = 0.050, p = 0.025) in order of de-
scending significance.
Goodman and Kruskal's Tau was chosen, as it offers advantages over all other meas-
ures of association for nominal variables. It can be interpreted easily, its maximum is one,
it takes the value 0, if the variables are independent, and it is suitable, if the number k of
the independent variable x's categories is not equal to the number m of the dependent vari-
able y's categories. Its only disadvantage is its complicated calculation, which is remedied
with modern statistics software.878 Goodman and Kruskal's Tau measures the association
of two categorical variables in terms of an improvement of prediction relative to a random
distribution of people (or responding companies in this case) based on the marginal distri-
bution.879 It is a measure of proportional reduction of error (PRE).880
The above Tau of 0.147, for instance, means that the postponement knowledge of
14.7 % more of responding companies is predicted correctly, if we predict the postpone-
ment knowledge of a company not only according to the general distribution of postpone-
ment knowledge, but according to the conditional distribution of postponement knowledge
that is distinguished by the risk pooling knowledge.881 Postponement might be the most
popular type of risk pooling, as its recognition shows the highest correlation with the
knowledge of risk pooling. Forty-two of the 59 responding companies who know risk pool-
ing also know postponement. Twenty-nine of the 43 logistics professionals not knowing
risk pooling do not know postponement either. The knowledge of risk pooling is also high-
ly significantly associated with knowing product pooling (τ = 0.104, p = 0.001) and trans-
shipments (τ = 0.091, p = 0.002).
Further significant correlations at least at the 5 % level between the knowledge of the
different risk pooling concepts can be read from table A.2 at the end of this section.
878
Müller-Benedict (2007: 203, 206).
879
Müller-Benedict (2007: 203).
880
Müller-Benedict (2007: 200).
881
Cf. Müller-Benedict (2007: 204).
151
Appendix A
Most significantly correlated is the knowledge of product pooling and transshipments

(τ = 0.163, p = 0.000), risk pooling and postponement (τ = 0.147, p = 0.000), transshipments
and capacity pooling (τ = 0.139, p = 0.000), as well as transshipments and postponement
(τ = 0.135, p = 0.000) and vice versa in order of descending strength of correlation.
In this sample the knowledge of postponement and capacity pooling (τ = 0.116,
p = 0.001), product substitution and demand reshape (τ = 0.112, p = 0.001), product pool-
ing and demand reshape (τ = 0.104, p = 0.001), postponement and order splitting
(τ = 0.101, p = 0.001), selective stocking and transshipments (τ = 0.098, p = 0.002), as well
as demand reshape and transshipments (τ = 0.095, p = 0.002) is relatively strongly corre-
lated. These concepts seem to be related and therefore to be known together: Postponement
and capacity pooling are popular and applied together e. g. in the auto industry. Both in
product substitution and demand reshape goods are substituted. One can offer a limited
number of products by product pooling or offer a variety of products (but not store all of
them) and apply demand reshape or transshipments. In order splitting certain steps in the
order and supplier delivery process may be postponed. If SKUs are only stocked at certain
locations (selective stocking), transshipments may be necessary between these locations.
Usually, we call a correlation in the social and economic sciences strong, if the measure
of correlation is > 0.5. In social and economic sciences any two characteristics are in gen-
eral only weakly connected, i. e. the measure of correlation is < 0.3 for a lot of bivariate
distributions of socio-scientific data, because social and economic interrelations are not
transparent at first glance, but multidimensional, flexible, and versatile.882 Furthermore, the
other nominally scaled measures of correlation show much higher values than Tau, but are
difficult to interpret except for Lambda. However, Lambda has the disadvantage of taking
the value 0, even if both variables are not independent.883 In our research the Tau values
are one half to one third of the other measures' ones.
A statistical correlation does not necessarily mean that there is a real cause and effect
relationship between two variables: First of all, with a 95 % confidence interval there is
still a 5 % chance that a measured correlation is coincidental. Secondly, without a theory
one cannot determine the cause and the effect. Thirdly, there might be a spurious causa-
tion, i. e. two characteristics change in the same way, although they are not connected.
They both are connected to and influenced by a third characteristic. In order to elucidate
these effects also statistically one has to analyze more than two variables collectively by
882
Müller-Benedict (2007: 197f.).
883
Müller-Benedict (2007: 203ff.).
152
Appendix A
multivariate data analysis.884 Multivariate analysis methods for only nominal data seem
less powerful and clear than the ones for metric data885 and are not widely available in
software packages886. If all variables are nominal, contingency table, configuration fre-
quency, log-linear, latent structure (if additional latent variables are used), and cluster
analysis can be employed. Most multivariate analysis methods require higher scales of
measurement.887 They should not be used without understanding them, their assumptions,
controversies, and theoretical foundations as modern software facilitate and some re-
searchers do.888
Knowledge of a risk pooling concept certainly is not only connected to the knowledge
of other risk pooling concepts and a company's industrial sector and size (number of em-
ployees), but also to other factors. Likewise, the application of a risk pooling concept
might be correlated to additional factors besides the awareness of this and other concepts,
the application of other concepts, and a company's sector of economic activity and size.
These other factors remain subject to further research.
884
Reynolds (1984: 72ff.), Müller-Benedict (2007: 267ff.).
885
Reynolds (1984: 78), Holtmann (2007).
886
Breakwell (2007: 441).
887
Holtmann (2007).
888
Breakwell (2007: 441f.).
153
Appendix A
Risk SRL PE Selective Commo‐ Turnover Product Product Demand Transship‐ Postpone‐ Virtual Capacity Central Order
Pooling Stocking nality Curve Pooling Substi‐ Reshape ments ment Pooling Pooling Ordering Splitting
tution
a
Risk 0.066 0.087 0.104 0.061 0.091 0.147 0.050
1 b
Pooling (0.010 ) (0.003) (0.001) (0.013) (0.002) (0.000) (0.025)
0.046 0.070 0.042 0.042 0.046 0.046
SRL 1
(0.031) (0.008) (0.040) (0.040) (0.032) (0.032)
0.067
PE 1
(0.009)
Selective 0.059 0.038 0.098 0.042
1
Stocking (0.015) (0.049) (0.002) (0.039)
Commo‐ 0.066 0.059 0.056 0.046 0.044 0.088
1
nality (0.010) (0.015) (0.017) (0.031) (0.034) (0.003)
Turnover 0.087 0.046 0.042 0.052 0.054 0.054 0.045 0.071
1
Curve (0.003) (0.031) (0.039) (0.022) (0.019) (0.019) (0.033) (0.007)
Product 0.104 0.056 0.052 0.104 0.163 0.058 0.054
1
Pooling (0.001) (0.017) (0.022) (0.001) (0.000) (0.016) (0.020)
Product
0.052 0.112 0.087
Substitu‐ 1
(0.022) (0.001) (0.003)
tion
Demand 0.061 0.038 0.042 0.104 0.112 0.095 0.071 0.044
1
Reshape (0.013) (0.049) (0.039) (0.001) (0.001) (0.002) (0.007) (0.035)
Trans‐
0.091 0.067 0.098 0.163 0.095 0.135 0.139
ship‐ 1
(0.002) (0.009) (0.002) (0.000) (0.002) (0.000) (0.000)
ments
Post‐
0.147 0.070 0.042 0.046 0.052 0.058 0.135 0.116 0.042 0.101
pone‐ 1
(0.000) (0.008) (0.039) (0.031) (0.022) (0.016) (0.000) (0.001) (0.040) (0.001)
ment
Virtual 0.050 0.042 0.054 0.087
1
Pooling (0.025) (0.040) (0.019) (0.003)
Capacity 0.042 0.044 0.054 0.139 0.116
1
Pooling (0.040) (0.034) (0.019) (0.000) (0.001)
Central 0.045 0.071 0.042 0.072
1
Ordering (0.033) (0.007) (0.040) (0.007)
Order 0.046 0.088 0.071 0.054 0.044 0.101 0.072
1
Splitting (0.032) (0.003) (0.007) (0.020) (0.035) (0.001) (0.007)
Table A.2: Significant Correlations at the 5 % Level of the Knowledge of Risk Pooling Con-
cepts
a. Goodman and Kruskal’s Tau

b. Approximative significance based on chi-square approximation.
2.2.3 Association between Knowledge and Utilization of Risk Pooling

Concepts
There is a strong highly significant correlation between the knowledge of a risk pooling con-
cept and its application as the diagonal in the correlation matrix in Table A.3 at the end of this
section shows. This suggests that increasing practitioners' knowledge and understanding of risk
pooling by praxis-oriented research may increase its application. The strongest correlation ex-
ists between the knowledge and the application of virtual pooling (τ = 0.431, p = 0.000). That
means the error of predicting the variable utilization of virtual pooling can be reduced by
43.1 %, if the distribution of the variable knowledge of virtual pooling is known. Forty-six of
the 65 respondents who know virtual pooling also apply it. Of the 37 respondents who do not
know virtual pooling 36 also do not apply it. Consequently, one professional did not recognize
the concept perhaps by its name, but when reading the explanation he came to realize that his
154
Appendix A
company applies this concept. There are a few cases like that pertaining to the other risk pool-
ing concepts as well.
The next highest correlations exist between the knowledge and application of the SRL
(τ = 0.319, p = 0.000) and of commonality (τ = 0.317, p = 0.000).
The recognition of transshipments is significantly correlated with both the application
of selective stocking (τ = 0.088, p = 0.003) and risk pooling (τ = 0.071, p = 0.008), the
knowledge of postponement with the use of risk pooling (τ = 0.079, p = 0.005). This
highlights again that the respondents seem to mainly associate the popular concepts of
postponement and transshipments with risk pooling. The knowledge of transshipments
might favor the implementation of selective stocking, since products have to be exchanged
between locations as they are only stored at few locations for economical reasons.
Utilization Risk SRL PE Selective Commo‐ Turnover Product Product Demand Trans‐ Post‐ Virtual Capacity Central Order Centra‐ Centra‐ Decentra‐ Decentra‐
Utilization Pooling Stocking nality Curve Pooling Substitu‐ Reshape ship‐ pone‐ Pooling Pooling Ordering Splitting lization lization lization lization
Knowledge tion ments ment past future past future
a
0.202
Risk Pooling b
(0.000 )
0.319 0.047
SRL
(0.000) (0.029)
0.291
PE
(0.000)
Selective 0.292
Stocking (0.000)
0.317
Commonality
(0.000)
Turnover 0.288 0.058 0.053 0.039
Curve (0.000) (0.016) (0.021) (0.048)
Product 0.098
Pooling (0.002)
Product 0.119 0.037
Substitution (0.001) (0.054)
Demand 0.041 0.151 0.041
Reshape (0.043) (0.000) (0.042)
Transship‐ 0.071 0.088 0.304
ments (0.008) (0.003) (0.000)
0.079 0.178 0.039
Postponement
(0.005) (0.000) (0.047)
0.045 0.431
Virtual Pooling
(0.032) (0.000)
Capacity 0.301
Pooling (0.000)
Central 0.128
Ordering (0.000)
0.141
Order Splitting
(0.000)
Table A.3: Significant Correlations at the 5 % Level (except in the case Product Substitu-
tion/Demand Reshape) between Knowledge and Utilization of Risk Pooling Concepts

155
Appendix A
2.2.4 Association of the Utilization of Different Risk Pooling Concepts
Most significantly and strongly associated is past decentralization with future decentralization
(τ = 0.253, p = 0.000), the utilization of product substitution and demand reshape (τ = 0.212,
p = 0.000), past centralization and decentralization (τ = 0.185, p = 0.000), the application of the
inventory turnover curve and transshipments (τ = 0.132, p = 0.000), commonality and product
pooling (τ = 0.127, p = 0.000), risk pooling and transshipments (τ = 0.121, p = 0.000), the
turnover curve and order splitting (τ = 0.117, p = 0.001), selective stocking and virtual pooling
(τ = 0.113, p = 0.001), transshipments and virtual pooling (τ = 0.113, p = 0.001), risk pooling
and postponement (τ = 0.109, p = 0.001), product substitution and transshipments (τ = 0.096,
p = 0.002), as well as product substitution and virtual pooling (τ = 0.096, p = 0.002) and vice
versa as shown in Table A.4 at the end of this section.
Eighty-four (99 %) of the 85 companies which did not decentralize in the past do not
intend to decentralize in the future. Eleven of the 17 companies which did decentralize will
continue to decentralize in the future. Eighty-four (88 %) of the 95 companies which do
not intend to decentralize their warehouse or logistics system, did not decentralize in the
past either. Six (86 %) of the seven companies planning to decentralize in the future, al-
ready decentralized in the past. It seems that these companies follow their warehouse or
logistics system strategy of decentralization or no decentralization consistently over time.
Fifty-one (94 %) of the 54 respondents who do not apply product substitution do not
apply demand reshape either. Twenty of the 48 respondents using product substitution
use demand reshape as well. Fifty-two of the 80 companies which do not reshape demand
do not substitute products in case of a stockout either. Twenty of the 22 demand reshaping
companies also use product substitution. It seems that companies which do not persuade
customers to buy a substitute product in case of a stockout mostly do not do so either, if the
desired product is available. This might occur, if a company does not sell substitute prod-
ucts. Ninety-one percent of demand reshapers also apply product substitution, perhaps be-
cause both strategies are similar in that they both substitute products.
If a company centralized in the past (49 companies), there was absolutely no decentra-
lization in the past. If a company conducted no centralization in the past, there was most-
ly no decentralization either: Of the 53 companies which did not centralize in the past only
17 decentralized in the past, 36 ones (68 %) did not. Past centralization and decentraliza-
tion thus show a rather negative correlation as the Phi value of -0.430 (p = 0.000) sug-
156
Appendix A
gests889. Forty-nine (58 %) of the 85 companies which did not decentralize centralized in
the past and none of the 17 companies which decentralized centralized in the past. It ap-
pears that decentralization excludes centralization and vice versa. The strategy of decentra-
lization, centralization, or no change is followed fairly consistently over time.
Thirty-four (97 %) of the 35 companies which do not transship do not use the inventory
turnover curve. Forty-three of the 67 transshipping companies utilize the turnover curve.
Thirty-four of the 77 responding companies which do not use the inventory turnover curve
do not take advantage of transshipments either. Twenty-four (96 %) of the 25 companies
using the turnover curve transship. One could assume that companies using the inventory
turnover curve attach great importance to inventory turnover. If therefore they tend to
mainly store fast movers at every location, this might explain why they also use transship-
ments to exchange e. g. slow movers between locations.
Forty (77 %) of the 52 companies which have not implemented component commonal-
ity do not use product pooling. Twenty-nine of the 50 companies using common compo-
nents also use product pooling. Forty of the 61 companies which have not implemented
product pooling have not implemented component commonality either. Twenty-nine
(71 %) of the 41 product pooling companies use common components as well. This comes
as no surprise as product pooling may rely on common components to satisfy customers
that were previously offered their own product variant with a single common product890.
Furthermore, both methods take advantage of standardization. “[S]tandardization enables
risk pooling across products, leading to lower inventories, and allows firms to use the in-
formation contained in aggregate forecasts more effectively”891.
Thirty-one (89 %) of the 35 companies not relying on transshipments likewise de-
clared that they did not apply risk pooling. Thirty-one of the 67 companies using trans-
shipments also use risk pooling. Thirty-six of the 67 companies which indicated that they
are not using risk pooling do not use transshipments either. Thirty-one (89 %) of the 35
ones using risk pooling do use transshipments as well. This highlights again that the re-
sponding professionals strongly associate risk pooling with transshipments.
Fifty-seven (74 %) of the 77 companies which do not use the turnover curve do not
apply order splitting either. Sixteen of the 25 companies employing the turnover curve
split orders as well. Fifty-seven (86 %) of the 66 companies not splitting orders do not use
889
Although we deal with nominal variables, we can exceptionally refer to a negative correlation here (cf.
Schulze 2007: 127).
890
891
Simchi-Levi et al. (2008: 357), cf. Reiner (2005: 434), Sheffi (2006, 2007: 186-193).
157
Appendix A
the turnover curve. Sixteen of the 36 companies that utilize order splitting use the invento-
ry turnover curve as well. Splitting an order into multiple orders or goods receipts may not
only decrease supplier lead time variability but also reduce average inventory and therefore
increase inventory turnover. The inventory turnover is reflected in the inventory turnover
curve. Inventory policies, which may be determined and controlled with the inventory
turnover curve, might lead to order splitting.
Twenty-seven of the 55 not virtually pooling companies do not store SKUs selectively
at warehouses. Thirty-nine (83 %) of the 47 companies which apply virtual pooling also
apply selective stocking. Twenty-five (71 %) of the 35 logistics professionals indicating
that their company does not rely on selective stocking do not pool virtually either. Thirty-
nine of the 67 companies using selective stock-keeping also apply virtual pooling. Virtual
pooling may entail selective stocking and vice versa: On the one hand, access to other loca-
tions' inventories via virtual pooling may favor a specialization in stock holding. On the
other hand, stocking products only at few locations due to cost considerations may necessi-
tate that these locations can access each other's inventories. For instance, a company may
choose to stock only fast movers at all regional warehouses and have slow movers drop-
shipped to customers or cross-filled. Thus inventories and demands at different locations
are pooled virtually.
Seventy-seven percent (27) of the companies not using transshipments (35), do not pool
virtually either. Fifty-eight percent (39) of the companies transshipping (67) use virtual
pooling in addition. Twenty-seven of the 55 responding companies which do not use vir-
tual pooling do not use transshipments either. However, thirty-nine (83 %) of the 47 vir-
tually pooling companies rely on transshipments. Extending a company's warehouse or
warehouses beyond its or their physical inventory to the inventory of other own locations
or of other companies' locations via information technology and the internet (virtual pool-
ing) often entails stock transfers between locations (transshipments). Virtual pooling re-
sembles transshipments, if it comprises drop-shipping892 or cross-filling893.
Of the 67 respondents who indicated not having implemented risk pooling 51 ones
(76 %) have not implemented postponement either. Twenty (57 %) of the 35 ones who
use risk pooling also use postponement. Fifty-one (77 %) of the 66 survey participants in-
dicating no postponement utilization also denied the use of risk pooling. Twenty (56 %) of
the 36 companies having implemented postponement also confirm the usage of risk pool-
892
893
158
Appendix A
ing. Consequently, the responding logisticians strongly associate risk pooling with post-
ponement.
Twenty-six (74 %) of the 35 companies that do not use transshipments do not substi-
tute products either. Thirty-nine of the 67 transshipping companies also use product sub-
stitution. Twenty-six of the 54 companies not substituting products do not use transship-
ments either. Thirty-nine (81 %) of the 48 product substitutionists use transshipments in
addition. This shows that product substitution and transshipments can be applied side by
side just like in the case of Papierco. For this company product substitution is the cheaper
option, so that it should only use transshipments as an alternative, if it is not possible to
substitute products in case of a stockout.
The same applies to virtual pooling: Thirty of the 48 companies which rely on product
substitution pool virtually as well. Thirty-seven (69 %) of the 54 ones not substituting
products, do not apply virtual pooling either. Thirty-seven (67 %) of the 55 companies not
pooling virtually do not use product substitution either. Thirty of the 47 ones pooling vir-
tually also substitute products.
form postponement at best only weakly for our sample: Postponement is only weakly and
at the 5 % level just not statistically significantly associated anymore with transshipments
(τ = 0.035, p = 0.059) and selective stocking (τ = 0.035, p = 0.059). Twenty-eight (42 %)
of the 67 companies stocking selectively thereby centralizing inventory also apply post-
ponement. Twenty-seven (77 %) of the 35 ones not using selective stocking do not use
postponement either. The same applies to transshipments.
Postponement is statistically significantly independent of past (τ = 0.000, p = 0.903)
and future centralization (τ = 0.009, p = 0.350). The contrary is true for its association with
risk pooling (τ = 0.109, p = 0.001) and component commonality (τ = 0.068, p = 0.009).
Forty of the 66 not postponing companies do not use component commonality. Twenty-
four (67 %) of the 36 postponing companies also employ common components. Forty
(77 %) of the 52 companies not applying component commonality do not apply postpone-
ment either. Twenty-four of the 50 users of common components also rely on postpone-
ment. This shows that the implementation of postponement and commonality may go hand
159
Appendix A
in hand894 as we already highlighted describing the Risk Pooling Decision Support Tool.
The implementation of postponement may even require component commonality.895
Risk SRL PE Selective Commo‐ Turnover Product Product Demand Trans‐ Post‐ Virtual Capacity Central Order Centra‐ Centra‐ Decentra‐ Decentra‐
Pooling Stocking nality Curve Pooling Substi‐ Reshape ship‐ pone‐ Pooling Pooling Ordering Splitting lization lization lization lization
tution ments ment past future past future
Risk a
0.066 0.068 0.052 0.121 0.109 0.059
Pooling 1 b
(0.010 ) (0.009) (0.022) (0.000) (0.001) (0.014)
SRL 0.066
1
(0.010)
PE 0.077 0.039 0.074 0.069
1
(0.005) (0.047) (0.006) (0.008)
Selective 0.051 0.068 0.113 0.045 0.038
1
Stocking (0.023) (0.009) (0.001) (0.033) (0.049)
Commo‐ 0.127 0.068
1
nality (0.000) (0.009)
Turnover 0.068 0.04 0.132 0.088 0.117
1
Curve (0.009) (0.045) (0.000) (0.003) (0.001)
Product 0.127 0.05
1
Pooling (0.000) (0.025)
Product
0.052 0.077 0.051 0.212 0.096 0.096
Substi‐ 1
(0.022) (0.005) (0.023) (0.000) (0.002) (0.002)
tution
Demand 0.039 0.04 0.212 0.079 0.077
1
Reshape (0.047) (0.045) (0.000) (0.005) (0.005)
Transship‐ 0.121 0.074 0.068 0.132 0.096 0.113
1
ments (0.000) (0.006) (0.009) (0.000) (0.002) (0.001)
Postpone‐ 0.109 0.068
1
ment (0.001) (0.009)
Virtual 0.059 0.113 0.088 0.096 0.079 0.113 0.079
1
Pooling (0.014) (0.001) (0.003) (0.002) (0.005) (0.001) (0.008)
Capacity 0.052
1
Pooling (0.022)
Central 0.052 0.058
1
Ordering (0.022) (0.016)
Order 0.069 0.117 0.079
1
Splitting (0.008) (0.001) (0.008)
Centra‐
0.058 0.185
lization 1
(0.016) (0.000)
past
Centra‐
lization 1
future
Decentra‐
0.045 0.05 0.077 0.185 0.253
lization 1
(0.033) (0.025) (0.005) (0.000) (0.000)
past
Decentra‐
0.038 0.253
lization 1
(0.049) (0.000)
future
Table A.4: Significant Correlations at the 5 % Level between the Utilization of Risk Pooling
Concepts

894
Zinn and Bowersox (1988: 124f.), Lee (1996, 1998), Van Hoek (1998b, 2001: 173), Van Hoek et al.
(1998), Feitzinger and Lee (1997), Pagh and Cooper (1998), Swaminathan and Tayur (1998), Aviv and
Federgruen (2001a, 2001b: 579), Swaminathan (2001), Yang et al. (2005b: 993), Simchi-Levi et al.
(2008: 345).
895
160
Appendix A
2.2.5 Risk Pooling Knowledge and Utilization in the Responding

Manufacturing and Trading Companies
Figure A.2: Risk Pooling Knowledge and Utilization in the Sample Manufacturing and
Trading Companies
161
Appendix A
When we subdivide the original sample for subanalyses in this section, we decrease our sample
size and therefore enlarge our confidence intervals896.
At the 5 % level Pearson's and Yates' chi-square and Fisher's exact test revealed signifi-
cant differences between manufacturing and trading companies in employing product sub-
stitution (Pearson's chi-square χ2 = 8.01, degrees of freedom df = 1, p = 0.004652), com-
monality (χ2 = 7.84, df = 1, p = 0.00511), and transshipments (χ2 = 6.19, df = 1,
p = 0.012847), in the knowledge of commonality (χ2 = 5.47, df = 1, p = 0.019346), and in
the utilization of postponement (χ2 = 4.52, df = 1, p = 0.033501897) in descending order of
asymptotic significance (two-sided).
The 75 manufacturing companies surveyed know central ordering, product substitution,
selective stocking, and order splitting the most, followed by transshipments, product pool-
ing, commonality, capacity pooling, virtual pooling, risk pooling, postponement, demand
reshape, the PE, inventory turnover curve, and SRL.
The 27 trading companies recognize central ordering, product substitution, selective
stocking, transshipments, order splitting, product pooling, the PE, virtual pooling, capacity
pooling, demand reshape, commonality, risk pooling, the inventory turnover curve, post-
ponement, and the SRL in decreasing order of awareness.
The manufacturing and trading companies do not differ statistically significantly at the
5 % level in their knowledge of the various risk pooling concepts, except for commonality.
Seventy-six percent of the manufacturing and 52 % of the trading companies are aware of
this concept.
Selective stocking (63 %), central ordering (61 %), transshipments (59 %), and com-
monality (57 %) are the concepts most frequently applied by the manufacturers. Capacity
pooling (47 %), centralization, virtual pooling, product pooling, postponement, product
substitution, order splitting, risk pooling, the PE, inventory turnover curve, demand re-
shape, the SRL, and decentralization follow in order of descending employment.
Trading companies apply transshipments (85 %), central ordering (78 %), selective
stocking (74 %), and product substitution (74 %) the most, followed by centralization
(63 %), virtual pooling, order splitting, risk pooling, the PE, product pooling, demand re-
shape, commonality, capacity pooling, the inventory turnover curve, postponement, decen-
tralization, and the SRL.
896
Vockell and Asher (1995: ch. 8).
897
However, in this case χY = 3.58 with df = 1 and p = 0.058479, but p = 0.037049 in Fisher's exact test.
162
Appendix A
All concepts are more widely used in the trading companies than in the manufacturing
companies, except for the SRL, commonality, the inventory turnover curve, product pool-
ing, postponement, and capacity pooling. As already mentioned above, this relationship is
statistically significant only for product substitution, commonality, transshipments, and
postponement at the 5 % level. This supports our hypothesis from the Risk Pooling Deci-
sion Support Tool that component commonality, postponement, and capacity pooling
are rather applied by manufacturers and might be more suitable in a manufacturing en-
vironment. However, the utilization of capacity pooling in manufacturing and trading
companies is just not statistically significantly different at the 5 % level anymore
(χ2 = 3.53, df = 1, p = 0.060268).
Forty-one percent of the manufacturing companies apply postponement. This matches
the more than 40 % of the manufacturers surveyed by Kisperska-Moroñ (2003: 131) in
Upper Silesia, Poland that gear total production towards customer orders.
Thirty-seven percent of the 27 electronics (classification numbers 30, 31, and 32) and
automotive (34) and 40 % of the five food (15) and clothing (18) manufacturers employ
postponement. Consequently, with our sample we cannot support at the 5 % level Van
Hoek (1998b), who finds that postponement is extensively applied in electronics and au-
tomotive and less extensively applied in food and clothing. The utilization of postpone-
ment by electronics and automotive and food and clothing manufacturers in our sample is
independent in the Fisher Exact Probability Test898 (p (two-tailed) = 1). It differs neither in
these two nor in the four groups significantly at the 5 % level. However, Van Hoek's
(1998b) observation may apply to the whole population, as the confidence interval for the
aforementioned conditions is very broad with 18.86 for electronics and automotive and
43.83 for food and clothing. For the same reasons as given earlier this confidence interval
can only serve as a crude estimate here. Only five answers from food and clothing manu-
facturers are probably not enough to compare postponement application between different
industrial sectors.
898
The Fisher Exact Probability Test is used, because some contingency table (crosstab) cells have expected
counts less than 5. Nevertheless, Pearson's and Yates' chi-square test led to the same conclusion.
163
Appendix A
2.2.6 Knowledge and Utilization of the Different Risk Pooling Concepts in

Small and Large Responding Companies
Figure A.3: Knowledge and Utilization of the Different Risk Pooling Concepts in Small and
Large Responding Companies
164
Appendix A
The answers of the 33 small (having less than 500 employees) and 69 large companies (having
500 or more employees according to the Deutsches Institut für Mittelstandsforschung (German
Institute for Midium-sized Businesses Research)) are independent (the null hypothesis that
there is no significant relationship between the answers cannot be rejected) in Pearson's and
Yates' chi-square and Fisher's exact test at the 5 % level, except for the usage of risk pooling
(χ2 = 5.63, df = 1, p = 0.017656), centralization in the past (χ2 = 4.75, df = 1, p = 0.029298),
and postponement (χ2 = 4.24, df = 1, p = 0.039482899). This means only for these last three
variables the distribution in the two groups of small and large companies differs statistically
significantly900, most significantly for risk pooling.
The small companies tend to have more knowledge of selective stocking, demand re-
shape, central ordering, product pooling, and order splitting, while a higher percentage of
large ones know the SRL and postponement. The large companies apply all concepts more
than small companies (especially risk pooling, postponement, transshipments, capacity
pooling, the PE, and product pooling) except for centralization in the past, demand re-
shape, commonality, product substitution, and central ordering in order of descending per-
centage difference of companies applying the respective concept. Hence, we can agree
with Huang and Li (2008b: 12) that large companies may apply postponement more than
small ones, since they may be better able to afford the redesign of products and/or
processes usually required by postponement901.
899
However, in this case χY = 3.37 with df = 1 and p = 0.066394, but p = 0.047623 in Fisher's exact test.
900
This means that one can be certain that this statistic is dependable, the difference is genuine and not by
coincidence. It does not mean that the difference is large, important, or useful (Walonick 2009).
901
Ernst and Kamrad (2000), Aviv and Federgruen (2001b).
165
3 Critical Appraisal
We need to point out that the assumption of the significance test of random sample data may
be violated. For population data any crosstable differences are real and thus significant. With
non-random sample data we cannot establish significance.902 However, significance tests are
sometimes used as rough approximations.903 For criticism of significance testing you may e. g.
refer to Carver (1978), Cohen (1994), and Thompson (1996). Statistical significance does not
tell if a research's results will reoccur904 or the strength and generalizability of relationships in a
sample and hence may divert attention from more crucial concerns905.
Although a thorough research process was followed, the limited representativeness of
this survey has to be acknowledged. This should be remedied in another survey by a higher
response rate. However, this might be difficult, as response rates for academic studies are
low906 and have been declining over the last years907. Perhaps there are so few surveys in
logistical research, since it is very difficult to obtain a random or at least representative
sample because of low response rates as we experienced. A lot of companies are swamped
with surveys and/or have a policy of not answering surveys. Some researchers pay compa-
nies to take part in surveys or give them other perquisites, which might also lead to a re-
sponse bias. Practitioners should not criticize business research as too theoretical on the
one hand and not take part in surveys on the other hand.
As it is laborious to design and administer surveys, the usually low response rates dis-
appoint the researcher and “imperil the strength of the research findings”908, and survey
research is prone to so many biases and errors909, one perhaps should consider other empir-
ical research. Netessine (2005: 6), for instance, suggests using university research data-
base, publicly available, consulting company, or industry journal data. Still for a Ph.D.
student with little funds it is difficult to access some of these data sources that are subject
to charges.
Besides, some specific data are not available at all or insufficient. For example, we
could not obtain any information about the number of warehouses from business and logis-
902
Garson (2009).
903
Dorofeev and Grant (2006: 253f.), Garson (2009).
904
Carver (1978).
905
Thompson (1994).
906
Bickard and Schmittlein (1999), Goldsby and Stank (2000).
907
Baruch (1999).
908
909
Groves et al. (2004: 39ff., 377f., 380).
166
Appendix A
tics associations and federal statistical offices in Germany and the U.S. to derive how in-
ventories changed in dependence of the number of warehouses compared to SRL predic-
tions or how the number of warehouses changed over time. Some researchers expected the
number of warehouses to increase with increasing fuel costs.
Furthermore, the number of companies in the industrial sectors in different publications
of the German Federal Statistical Office differs, as it receives its information from different
data sources. If the industrial sectors are broken down further, only companies with 20 or
more employees are considered.910 If all companies with any number of employees are
considered, they are not subdivided into detailed industrial sectors.911 This impaired our
retrospective quota sampling.
Another survey could also shed light onto the companies' reasons for and against as well
as the manner, degree, benefits, and costs of applying the risk pooling concepts, as we in-
tended in the six-page version of this survey that yielded a uselessly low response rate. As
risk pooling can affect the whole supply chain, another survey could consider a supply
chain wide perspective rather than single companies.
A lot of respondents do not perceive all the questioned concepts as risk pooling ones. It
appears that mainly transshipments and postponement are associated with risk pooling.
Although at least in our sample the risk pooling concepts are known fairly well (central
ordering, product substitution, and selective stocking the most), only selective stocking,
transshipments, and central ordering are widely applied. Transshipments912 are more im-
portant and applied, postponement913 and product pooling914 less than suggested by some
publications mostly for other regions.
Nearly half the sample centralized its warehouse or logistics system in the past. This is
less than overall in Europe915. In contrast to the expected trend to decentralize due to in-
creasing transportation costs, only seven percent will increase, six percent even decrease
the number of warehouse or production locations in the future. It seems that the companies
are very consistent in their strategy of centralization, decentralization, or no change over
time.
910
For example Statistisches Bundesamt Deutschland (2009i: 371).
911
For example Statistisches Bundesamt Deutschland (2009i: 378).
912
CLM (1995), Herer et al. (2006).
913
CLM (1995: 209), McKinnon and Forster (2000: 5), Chiou et al. (2002), Alfaro and Corbett (2003: 12),
Oracle (2004), Huang and Li (2008b).
914
915
CLM (1995).
167
Appendix A
The utilization and knowledge of a risk pooling concept are correlated. This suggests
that improving the knowledge about risk pooling may increase its application. Research
needs to convey to practitioners under what conditions and how the different risk pooling
concepts can be applied successfully. This research constitutes one step in this direction.
Prominently associated is past decentralization with future decentralization, the utiliza-
tion of product substitution and demand reshape, past centralization and decentralization
(the only negative correlation), the application of the inventory turnover curve and trans-
shipments, commonality and product pooling, risk pooling and transshipments, the inven-
tory turnover curve and order splitting, selective stocking and virtual pooling, transship-
ments and virtual pooling, risk pooling and postponement, product substitution and trans-
shipments, product substitution and virtual pooling, as well as postponement and commo-
nality and vice versa.
form postponement at best only weakly for our sample.
Trading companies seem to apply transshipments and product substitution more than
manufacturing companies. The opposite is true for commonality and postponement. In
contrast to Van Hoek (1998b) our sample shows no significant difference in the application
of postponement by electronics, automotive, food, and clothing manufacturers.
More large companies appear to use risk pooling and postponement than smaller ones,
perhaps because they can afford to invest in expensive risk pooling strategies as suggested
by Huang and Li (2008b: 12). More small companies have centralized their warehouse or
logistics system in the past than large ones in our sample.
Our survey shows that different risk pooling methods are and can be applied together as
we already suggested in the Risk Pooling Decision Support Tool and the Papierco example.
168
4 Questionnaire
Survey on Risk Pooling in Business LogLVWLFV
by Dipl.-Kfm. Gerald Oeser
1 Line of business
2 Number of employees in Germany
3 Your position (e. g. logistics manager)
Do you know the following concepts? Does your company use them? Known? In
use?
4 Risk Pooling (Demand and/or lead time variability is reduced, if one aggregates demands and/ Yes Yes
or lead times e. g. across locations, because it becomes more likely that higher-than-average de- No No
mands and/or lead times will be offset by lower-than-average ones. This reduction in variability allows
to reduce safety stock and therefore reduces average inventor y without reducing the service level.)
5 Square Root Law (Total (safety) stock in n warehouses = (safet y) stock in 1 centralized Yes Yes
warehouse * square root of the number of warehouses n.) No No
6 The Portfolio Effect shows the reduction in aggregate safety stock by consolidating several ware- Yes Yes
houses' inventories in one warehouse in percent. No No
7 Selective Stock Keeping/Specialization (Reducin g inventory carr ying cost treating products
Yes Yes
differently without reducing the service level substant ially. For example, products with a low turn-
RYHU PLJKWRQO\EHVWRFNHGDW IHZORFDWLRQVGXH WRFRVWFRQV LGHUDWLRQV No No
8 Component Commonality (Products are designed sharing com ponents. This reduces the va- Yes Yes
riability of demand.) No No
9 The Inventory Turnover Curve shows average inventory in dependence of the inventory Yes Yes
throughput for a specific company. It can be used to estimate the average inventory for any
No No
(planned) warehouse throughput (shipments from the warehouse or sa les).
10 Product Pooling (Unification of several prod uct designs to a single generic or universal design Yes Yes
or reducing the number of products thereby serving dem ands that were previously served by their No No
own product version with fewer products.)
11 Product Substitution (One tries to make custom ers buy another alternative product, becauseYes Yes
the original customer wish is out of stock No No
12 RU DOWKRXJKWKHRULJLQDOFXVWRPHUZLVKLVDYDLODEOH'HPDQG 5HVKDSH 7KLV UHGXFHV YDULDELOL Yes Yes
ty of demand and inventory costs .) No No
13 Transshipments/Inventory Sharing Inventor y transfers among locations (e. g. between Yes Yes
warehouses or stores) for example in case of a stock out. No No
14 Delayed Product Differentiation and Postpon ement 3RVWSRQHPHQWRIHJPDQXIDFWXULQJ
DVVHPEO\SDFNDJLQJODEHOLQJRUGHOLYHU\3URGXFWFXVWRPL]DWLRQVKRXOGEHGHOD\HGDVORQJDVSRVVLYes Yes
EOH 7KLV DOORZV WR VKLS D VLQJOH JHQHULF SURGXFW IXUWKHU GRZQ WKH VXSSO\FKDLQDQGWRFKDQJHLW No No
ODWHULQWRLQGLYLGXDOSURGXFWV DFFRUGLQJ WR UHFHQW GHP DQG LQIRUP DWLRQ 'HP DQG IRU LQGLYLGXDOSUR
GXFWV LV DJJUHJDWHG WRWKH GHPDQG IRU WKH JHQHULF SURGXFWRQSUHFHGLQJVXSSO\FKDLQVWDJHV
$JJUHJDWHGGHPDQGIRUWKHJHQHULFSURGXFWIOXFWXDWHVOHVV VLQFHVWRFKDVWLFIOXFWXDWLRQVRIWKHLQGL
YLGXDOGHPDQGVDUHRIIVHWWRDFHUWDLQH[WHQW,QYHQWRU\FDUU\LQJFRVWVDQGVWRFNRXWVFDQEHUHGXFHG
15 Virtual Inventory/Warehouse/Pooling: $ FRPSDQ\ H[WHQGV LWVZDUHKRXVHVEH\RQGWKH Yes Yes
SK\VLFDO LQYHQWRU\ WR WKH LQYHQWRU\ RI RWKHU RZQ ORFDWLRQV RU RI RWKHU FRP SDQLHV
ORFDWLRQVXVLQJ No No
LQIRUPDWLRQDQGFRPPXQLFDWLRQWHFKQRORJ LHVGURSVKLSSLQJDQGFURVVILOOLQJ
16 Capacity Pooling &RQVROLGDWLRQRISURGXFWLRQVHUYLFHRUZDUHKRXVHFDSDFLWLHVRIVHYHUDO Yes Yes
IDFLOLWLHV:LWKRXW SRROLQJ HYHU\ IDFLOLW\ IXOILOOV GHP DQG MXVW ZLWK LWV RZQ FDSDFLW\:LWKSRROLQJ No No
GHPDQGLVDJJUHJDWHGDQG IXOILOOHG E\ D VLQJOH SHUKDSV YLUWXDOO\ MRLQW IDFLOLW\$KLJKHUVHUYLFH
OHYHOFDQEHDWWDLQHGZLWKWKHVDPHFDSDFLW\RU WKH VDP H VHUYLFH OHYHO FDQ EH RIIHUHGZLWKOHVV
FDSDFLW\
17 Central Ordering for several locations and later allocation of the orders (perhaps by a depot) Yes Yes
to the distribution points or requisitioners according to recent demand inform ation. No No
18 Order 6plitting LVSDUWLWLRQLQJDUHSOHQLVKPHQWRUGHULQWRPXOWLSOHRUGHUVZLWKPXOWLSOHVXSSOLHUVRU Yes Yes
LQWRPXOWLSOHGHOLYHULHV No No
19 a) Did you centralize your warehouse/logistics system (reduce the number of warehouse or Yes No
production locations)?
b) Do you intend to centralize your warehouse/logistics system? Yes No
20 a) Did you decentralize your warehouse/logistics system (increase the number of warehouse Yes No
or production locations)?
b) Do you intend to decentralize your warehouse/logistics system? Yes No
Please push to send answers
Figure A.4: Questionnaire

169
5 Answers
KNOWLEDGE UTILIZATION
Demand Reshape
Demand Reshape
Decentralization
Decentralization
Central Ordering
Central Ordering
Capacity Pooling
Capacity Pooling
Product Pooling
Product Pooling
Transshipments
Transshipments
Turnover Curve
Turnover Curve
Virtual Pooling
Virtual Pooling
Postponement
Postponement
Order Splitting
Order Splitting
Centralization
Centralization
Commonality
Commonality
classification
Substitution
Substitution
Respondent
Risk Pooling
Risk Pooling
Employees
Selective
Selective
Industry
Product
Stocking
Stocking
Product
future
future
past
past
SRL
SRL
PE
PE
1 52.42 16,000 1 1 1 1 0 1 1 1 1 1 1 1 0 1 1 1 0 0 1 0 1 0 1 1 1 0 1 0 1 0 0 0 1 1
2 52.42 5 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 0 0 1 1 1 0 0 1 0 1 0 1 0 1 0 1 0 0 0
3 29 2,000 1 0 1 1 1 0 1 1 1 1 1 1 0 1 1 1 0 1 1 1 0 0 1 0 1 1 1 0 1 0 1 0 0 0
4 29 3,000 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0 1 1 0 1 1 1 1 1 1 1 0 0 0
5 29 100 1 0 1 1 1 0 1 1 1 1 0 0 0 1 1 0 0 0 0 1 0 1 0 0 0 0 0 0 1 0 1 0 0 0
6 34 60 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 0 0 0 0 1 1 0 1 0 1 0 0 1 0
7 25 190 0 0 0 1 1 0 1 1 0 1 0 0 0 1 0 0 0 0 0 1 0 0 0 0 1 0 0 0 1 0 0 0 0 0
8 31 2,000 1 0 1 1 1 1 1 0 0 1 1 1 1 1 1 0 0 0 1 1 1 1 0 0 1 1 1 1 1 1 1 0 0 0
9 31 50 0 0 0 1 1 0 1 1 1 1 0 0 0 1 1 0 0 0 1 1 0 1 0 0 0 0 0 0 1 0 1 0 0 1
10 37 120 1 0 1 1 1 1 1 1 1 0 0 1 0 1 1 0 0 0 1 0 1 0 0 1 0 0 1 0 1 1 1 0 0 0
11 33 133,000 0 0 1 0 0 1 1 1 1 1 0 1 0 1 1 0 0 1 0 0 1 1 0 0 1 0 1 0 0 1 0 0 0 0
12 34 78,000 0 0 1 1 0 0 0 1 0 1 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
13 32 500 0 0 0 0 1 0 1 1 0 0 0 1 1 1 1 0 0 0 0 1 0 1 0 0 0 0 0 0 0 1 0 0 0 0
14 29 150 0 0 0 1 1 0 1 0 0 1 1 1 1 1 1 0 0 0 1 1 0 1 0 0 1 0 1 1 1 0 1 0 0 0
15 52.42 100 1 0 1 1 1 0 1 1 0 1 1 1 1 1 1 1 0 1 1 0 0 0 0 1 1 0 1 1 1 0 1 0 0 0
16 30 1,600 0 1 0 0 0 1 1 1 0 0 1 0 1 1 1 0 0 0 0 0 0 1 0 0 0 0 0 1 1 0 0 0 0 0
17 34 13,000 0 1 0 1 1 0 0 1 0 0 0 0 0 1 1 0 0 0 0 1 0 1 0 0 0 0 0 1 0 1 0 0 0 0
18 37 9,000 0 0 0 1 0 0 0 1 0 0 0 1 1 0 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0
19 52.11 121,000 0 0 0 1 0 0 0 1 0 1 0 1 1 1 0 0 0 0 1 0 0 0 1 0 1 0 1 1 1 0 0 0 1 1
20 51.46 30 1 0 0 0 0 1 0 1 1 0 0 1 0 1 1 0 0 0 0 0 0 0 1 1 0 0 1 0 1 1 1 0 0 0
21 45.2 120 0 0 1 1 1 0 0 0 0 0 0 0 1 1 0 0 0 1 1 1 0 0 0 0 0 0 0 1 1 0 0 0 0 0
22 25 143 1 0 1 1 0 0 1 1 1 1 1 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 0 0
23 35 1,500 0 0 0 0 0 0 0 0 0 1 1 0 0 1 0 0 0 0 0 0 0 0 0 0 1 1 0 0 1 0 0 0 0 0
24 29 47,296 1 1 1 1 1 0 1 1 0 1 1 1 1 1 0 1 1 0 1 1 0 0 0 0 1 1 0 0 0 0 0 0 0 0
25 31 1,000 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 0 1 0 0 0 0 0 0 1 1 0 0 1 0
26 52.11 170,000 0 0 1 0 0 0 0 1 0 0 0 0 0 1 1 0 0 1 0 0 0 0 1 0 1 0 0 0 1 0 1 0 0 0
27 52.42 20 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 0 0 0 1 1 1 1 0 0 1 0 1 1 1 1 0 0 0 0
28 52.11 150 0 0 1 1 1 1 1 1 1 1 0 1 0 1 1 0 0 1 1 0 0 1 1 0 1 0 1 0 0 1 1 0 0 0
29 52.11 65 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 1 0 1 1 1 0 1 0 1 0 0 1 0 0
30 28 346 1 0 0 1 1 1 1 1 1 1 0 1 1 1 1 1 0 0 1 1 1 1 1 1 1 0 1 0 0 0 0 0 1 0
31 16 2,500 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 0 0 0 1 0 1 0 0 0 1 0 1 0 0 1 0 0 0 0
32 15 500 1 1 0 1 1 0 1 1 1 1 1 0 1 1 1 1 0 0 0 0 0 1 0 0 0 1 0 1 0 0 0 0 1 0
33 34 2,000 0 0 1 0 0 0 1 1 0 1 1 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
34 29 600 0 1 0 0 0 1 0 1 0 0 0 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0 1 1 0 0 1 1
35 52.42 12,000 0 1 1 1 1 0 1 1 1 1 1 0 1 1 1 0 1 1 0 0 0 0 1 1 1 0 0 0 1 1 0 0 0 0
36 51.53.1 22 0 0 0 1 0 0 1 1 0 1 0 0 1 1 1 0 0 0 1 0 0 0 1 0 0 0 0 0 1 0 1 0 0 0
37 51.46 20 0 0 1 1 0 0 1 1 1 0 0 0 0 1 1 0 0 0 0 0 0 0 1 1 0 0 0 0 1 1 1 0 0 0
38 29 114,000 1 0 1 1 1 0 1 1 0 1 0 1 0 0 0 1 0 0 1 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0
39 15 800 1 1 0 1 1 1 0 1 0 1 1 0 1 1 1 0 0 0 1 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0
40 21 100 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 0 0 1 1 1 1 1 0 0 0 1 0 0 0
41 52.42 7,000 0 0 0 1 0 1 1 1 0 1 0 0 1 1 1 0 0 0 1 0 1 0 1 0 1 0 0 1 1 1 0 1 0 0
42 34 10,000 0 0 0 0 1 0 0 1 0 0 0 1 1 1 1 0 0 0 0 1 0 0 0 0 0 0 1 1 1 1 0 0 0 0
43 34 25,103 1 0 0 1 1 0 1 1 1 1 1 0 0 1 1 1 0 0 0 1 0 0 0 0 1 1 0 0 0 0 0 0 0 0
44 37 500 1 0 0 1 1 1 1 0 0 1 0 1 1 1 0 0 0 0 1 0 1 0 0 0 1 0 1 1 0 0 0 0 1 0
45 36 120 1 0 0 1 1 0 1 0 0 0 1 0 0 1 1 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0
46 34 12,000 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0 0 0
47 15 2,000 1 1 0 1 0 0 0 1 0 0 0 1 0 1 0 1 1 0 1 0 0 0 1 0 0 0 1 0 1 0 1 0 0 0
48 29 36,000 0 0 0 1 0 0 0 1 0 0 1 1 0 1 0 0 0 0 1 0 0 0 1 0 0 1 1 0 1 0 0 0 0 0
49 32 12,600 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 1 1 1 0 0 1 0 0 1 1 0 0 0
50 52.11 15,000 0 0 1 1 1 0 0 1 1 1 0 1 0 1 1 1 0 1 1 1 0 0 1 0 1 0 0 0 1 1 0 0 0 0
51 31 200 1 1 0 0 0 0 1 1 0 1 0 1 0 0 1 0 1 0 0 0 1 0 1 0 1 0 1 1 0 1 1 0 0 0
52 31 550 0 0 1 1 0 1 0 0 0 1 1 0 1 1 1 1 1 1 1 0 1 0 0 0 1 1 0 1 1 1 1 0 0 0
53 34 5,000 0 0 0 1 0 0 0 1 1 1 0 1 1 0 0 1 1 0 1 1 0 0 1 1 1 1 1 1 1 0 1 0 0 0
54 34 2,000 1 1 0 1 1 1 1 1 0 1 0 1 1 1 1 1 1 0 1 1 1 1 1 0 1 0 1 1 1 1 0 0 1 0
55 52.11 73,000 1 0 0 1 0 1 1 1 1 0 0 1 0 1 0 0 0 0 0 0 0 1 0 0 1 0 0 0 1 0 1 0 0 0
56 35 8,000 0 0 0 0 1 0 0 0 0 0 0 0 0 1 1 0 0 0 0 1 0 0 0 0 0 0 0 0 1 1 0 0 0 0
57 34 95,500 0 0 1 1 1 0 1 1 0 1 1 1 1 1 1 0 0 1 1 1 0 0 0 0 1 0 1 1 1 1 0 0 0 0
58 52.42 10,000 1 0 1 1 0 0 1 1 1 1 0 1 0 1 0 1 0 1 1 0 0 1 1 1 1 0 1 0 1 0 0 0 1 0
59 35 3,000 1 0 1 1 1 1 1 1 0 1 1 0 0 1 0 0 0 0 1 0 0 0 0 0 1 0 0 0 1 0 1 0 0 0
60 52.46 14,000 1 1 0 1 1 0 1 0 1 1 1 0 0 1 1 0 0 0 1 0 0 0 0 0 1 0 0 0 1 1 0 0 0 0
61 33 100 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 0 1 0 0 1 1 0 0 0 0 0 0 1 0
62 24 50 1 0 0 1 1 1 0 0 0 0 1 0 1 1 1 0 0 0 0 1 0 0 0 0 0 1 0 1 1 0 1 0 0 0
63 28 900 1 0 0 1 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 1 0 0 0
64 40 66,000 1 1 1 1 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 0 1 1 0 1 0 0 0
65 27 10 0 0 1 1 0 1 1 1 1 1 1 1 0 1 1 0 0 0 1 0 1 0 1 0 1 0 0 0 1 0 1 0 0 0
66 31 150 0 0 0 0 1 0 1 1 0 0 0 1 0 1 0 0 0 0 0 1 0 1 1 0 0 0 0 0 0 0 1 0 0 0
67 28 150 1 0 1 1 1 0 1 1 1 1 0 1 0 1 1 0 0 1 1 0 0 0 1 1 1 0 0 0 1 0 0 0 0 0
68 34 27,000 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 0 1 0 0 0 1 1 1 1 1 1 0 0 0
69 34 37,000 0 0 1 1 1 0 0 0 0 0 0 0 0 1 1 0 0 1 1 1 0 1 0 0 0 1 0 0 1 0 0 0 0 0
70 52.42 29,000 1 0 0 1 1 1 1 1 1 1 1 0 1 1 1 1 0 0 1 0 1 0 0 0 1 1 0 0 1 0 1 0 0 0
71 22 850 0 0 0 1 1 0 1 1 1 0 0 0 0 1 1 0 0 0 1 0 0 1 0 0 0 1 0 0 1 0 1 0 0 0
72 26 1,000 1 0 1 0 0 1 1 1 1 1 0 0 1 1 0 1 0 1 0 0 1 1 1 0 1 0 0 1 1 0 1 0 0 0
73 31 20 1 0 1 1 1 0 1 1 1 1 0 0 1 1 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0
74 28 1,200 0 0 1 1 1 0 1 1 0 0 0 1 0 1 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0
75 11 7,000 0 0 0 1 0 0 1 1 1 1 0 0 1 1 0 0 0 0 1 0 0 1 1 1 1 1 0 1 1 0 0 0 1 0
76 29 50,000 1 0 0 1 1 0 1 1 0 1 1 0 1 1 1 1 0 0 1 1 0 1 1 0 1 1 0 1 1 0 0 0 0 0
77 51.53.1 9,500 0 0 1 1 1 1 1 1 1 1 0 0 0 1 1 0 0 1 0 1 0 1 1 0 1 0 0 0 0 1 1 0 0 0
78 52.48.6 110 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 1 1 1 1 1 0 0 1 0 0 1 1
79 24 60,000 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 1 0 1 1 1 1 1 1 1 0 1 1 0 0 0 1 0
80 52.11 50 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 0 1 0 1 1 1 0 0 0
81 29 850 0 0 1 1 1 0 1 1 1 1 1 0 1 1 1 0 0 0 1 1 0 1 0 0 0 0 0 1 1 0 0 0 1 1
82 29 12,800 1 1 1 0 1 0 1 1 0 1 0 1 1 0 1 0 1 1 0 1 0 1 0 0 1 0 1 0 0 1 1 0 0 0
83 18 1,980 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 1 0 1 1 1 1 1 1 1 1 1 0 0 0
84 15 3,000 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 0 0 0 1 0 1 1 1 1 1 0 0 0
85 24 10,767 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 0 0
86 52.42 6,342 1 0 0 1 0 0 1 0 0 1 1 0 0 0 0 1 0 0 1 0 0 1 0 0 1 1 0 0 0 0 0 0 0 0
87 36 242 1 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 1 1 0 1 0 0 0
88 29 107 1 0 1 1 1 1 1 1 1 1 0 0 1 1 1 0 0 0 0 1 0 1 0 0 0 0 0 1 1 0 1 0 0 0
89 51.88 40 0 0 1 1 0 0 0 1 0 1 0 1 1 1 1 0 0 0 0 0 0 0 0 0 1 0 1 0 1 0 1 0 0 0
90 51.47.8 65 0 0 0 1 0 0 1 1 0 1 0 0 1 1 1 0 0 0 1 0 0 0 1 0 1 0 0 1 1 0 1 0 0 0
91 21 5,700 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 0 0 1 1 1 1 0 1 1 0 1 0 0 0
92 22 3,200 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 0 0 1 1 1 0 1 1 0 0 0 1 0 0
93 52.42 956 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0 1 1 1 0 1 1 0 1 1 1 1 1 0 1 0 0 0
94 29 1,000 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 0 0
95 34 60,000 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 0 0 1 1 0 0 1 1 1 1 0 1 0 1 0 0 1 1 0
96 40 1,850 1 0 0 1 1 0 1 1 0 1 1 1 1 0 1 1 0 0 1 0 0 0 1 0 1 0 1 1 0 0 0 0 0 0
97 32 60,000 1 0 1 1 1 1 1 1 0 1 0 1 0 0 1 0 0 0 1 1 0 1 1 0 1 0 1 0 0 0 1 0 0 0
98 32 6,000 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1 1 0 0
99 32 4,500 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 1 0 0 1 1 1 0 0 0 1 0 0 0
100 51.47.8 9 1 0 1 1 0 1 1 1 0 1 1 1 1 1 1 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0
101 34 182,739 1 0 0 1 1 0 1 1 1 1 1 1 1 1 1 0 0 0 1 1 0 0 0 0 1 1 0 1 0 0 0 0 0 0
102 51.47.8 1,652 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 1 0 1 1 1 0 0 0 0 0 1 1
1 = Yes, 0 = No
170
Appendix A
Industry Classification Description
11 Extraction of crude petroleum and natural gas; service activities incidental to oil and gas
extraction, excluding surveying
15 Manufacture of food products and beverages
16 Manufacture of tobacco products
18 Manufacture of wearing apparel; dressing and dyeing of fur
21 Manufacture of pulp, paper and paper products
22 Publishing, printing and reproduction of recorded media
24 Manufacture of chemicals and chemical products
25 Manufacture of rubber and plastic products
26 Manufacture of other non‐metallic mineral products
27 Manufacture of basic metals
28 Manufacture of fabricated metal products, except machinery and equipment
29 Manufacture of machinery and equipment n.e.c.
30 Manufacture of office machinery and computers
31 Manufacture of electrical machinery and apparatus n.e.c.
32 Manufacture of radio, television and communication equipment and apparatus
33 Manufacture of medical, precision and optical instruments, watches and clocks
34 Manufacture of motor vehicles, trailers and semi‐trailers
35 Manufacture of other transport equipment
36 Manufacture of furniture; manufacturing n.e.c.
37 Recycling
40 Electricity, gas, steam and hot water supply
45.2 Building of complete constructions or parts thereof; civil engineering
51.46 Wholesale of pharmaceutical goods
51.47.8 Wholesale of paper and paperboard, stationery, books, newspapers, journals and periodicals
51.53.1 Non‐specialized wholesale of wood, construction materials and sanitary equipment
51.88 Wholesale of agricultural machinery and accessories and implements, including tractors
52.11 Retail sale in non‐specialized stores with food, beverages or tobacco predominating
52.42 Retail sale of clothing
52.46 Retail sale of hardware, paints and glass
52.48.6 Retail sale of games and toys
Table A.5: Survey Participants' Answers to the Questionnaire
171
Appendix B: Proof of the Square Root Law for Regular,
Safety, and Total Stock
Based on a numerical example in Ballou (2004b: 379f.) and in a manner different from the
flawed916 one in Maister (1976), we formally prove that under certain assumptions the SRL
applies to regular, safety, and total stock.
Let average regular stock at location i be the economic order quantity (EOQ) Qi divided
by two:
, (B1)
where di is demand at warehouse i, S the fixed order cost, I the inventory carrying cost rate and
C the product value per unit. Regular system inventory for n decentralized warehouses equals
∑ ∑
. (B2)
Regular stock in one centralized warehouse is
916
Das (1978: 334) correctly remarks “But Maister's proof is incorrect due to a combination of faulty algebra
and possible misprints” without pointing out the mistakes. Indeed, we found e. g. that it is supposed to be
... ∑ instead of ... ∑ (Maister 1976: 129) and
∑ ∑ ∑
instead of ∑ (Maister 1976: 130). The expression ∑ (Maister 1976: 130) is also in-
∑
∑
correct. Furthermore, it has to be 0.2 0.005 instead of 0.02 0.005 (Maister 1976: 129).
Changing a three-location to a one-location system reduces inventory by 42 (not 43) % and changing a
ten-location to a four-location system reduces inventory by 37 (not 38) % according to the SRL (Maister
∑
1976: 131). Maister (1976: 132) wrongly states ∑ √ , although he ear-
lier notes that ∑ 1 (Maister 1976: 127). He also explains “If the original decentralised system had 16
locations, and inventory costs account for one-half of total costs in this system, then inventory savings

from centralisation equal 1 . But transportation must represent 1 , or
√
in the decentralised system. Hence, we can afford to almost double transportation expenses and still retain
a system cost less than in the original system” (Maister 1976: 133). However, we can only afford 1.75
times and not almost double the transportation cost or, in other words, we could only increase transporta-
tion cost by 75 % at the most. But if all inventory savings were spent on increased transportation costs,
we would not retain a system cost less than in the original system. The centralized and decentralized sys-
tem costs would be equal and we would be indifferent between them.
Nevertheless, Das (1978: 333) also misprinted 1 and incorrectly uses the equality instead of the ap-
proximation sign, although the power (square root) function cannot equal a linear one. Furthermore,
, ,..., should be , , . . . , and ∑ should be ∑ (Das 1978: 334). Das (1978: 332) also
refers to a proof to his solution in the appendix, but the journal does not contain it. Professor emeritus
Chandrasekhar Das, College of Business Administration, University of Northern Iowa, Department of
Management, agreed with us and kindly rewrote the proof, as it was not available at the publisher any-
more in 2009 (personal correspondence on February 27, 2009).
172
Appendix B
∑
∑ (B3)
.
Regular stock in one centralized warehouse times a factor α equals regular system stock
in n decentralized warehouses:
∑
∑ ∑
· ·
∑
(B4)
∑ ∑
· .
∑ ∑
If demand at each warehouse is equal (d), the regular system inventory for n decentra-
lized warehouses equals the one in one centralized warehouse times the square root of the
number of warehouses n:
∑ √ √ √ √ √
· · ·
∑ √ √ √ (B5)
·√ .
This means the SRL applies to regular stock, if an EOQ order policy is followed, the
fixed cost per order and the per unit stock holding cost (inventory carrying cost rate times
the product value per unit), demand at every location, and total system demand is the same
both before and after centralization.917 However, according to Maister (1976: 128f.) the
SRL can also be “a good approximation when demand rates at all the field locations are not
equal”.
If regular inventory in each of n locations in the decentralized system (RSs) is consi-
dered the following relationship holds:
·√ ·√
√ . (B6)
√ √ √
This corresponds to equation 9-28 in Ballou (2004b: 380): √ .

Safety stock in warehouse i is
√ , (B7)
where z is the same safety factor for every warehouse, sdi the standard deviation of demand at
warehouse i, and LT the constant replenishment lead time for every warehouse.
System safety stock in n decentralized warehouses is
√ ∑ . (B8)
917
Cf. Maister (1976: 125, 127, 131).
173
Appendix B
If the demands are uncorrelated, safety stock in the centralized warehouse is

√ ∑ . (B9)
Safety stock in one centralized warehouse times a factor α equals systemwide safety
stock in n decentralized warehouses:
· √ ∑ · √ ∑
√ ∑ ∑ ∑ (B10)
· .
√ ∑ ∑ ∑
If the standard deviation of demand at each warehouse is equal (sd), the systemwide
safety stock in n decentralized warehouses is equal to the safety stock in one centralized
warehouse times the square root of the number of warehouses n:
∑
· · ·
∑ √
(B11)
√ √
· ·√ .
√
This means the SRL applies to safety stock, if demands at the decentralized locations
are uncorrelated918, the variability (standard deviation) of demand at each decentralized
location919, the safety factor (safety stock multiple)920 and average lead time are the same
at all locations both before and after consolidation, average total system demand remains
the same after consolidation921, no transshipments occur922, lead times and demands are
independent and identically distributed random variables and independent of each other923,
the variances of lead time are zero924, and the safety-factor (kσ) approach is used to set
safety stock for all facilities both before and after consolidation925.
If demands at the warehouses are correlated, the SRL does not even yield an approxima-
tion. The safety stock savings through centralization decrease with increasing correlation
between demands at the warehouses and become zero with perfectly positive correlation.926
Total inventory is the sum of regular and safety stock.927 Therefore total inventory in the
decentralized system TId with n warehouses equals total inventory in one centralized ware-
house TIc times the square root of the number of warehouses n:
918
Maister (1976: 132).
919
Maister (1976: 130).
920
Maister (1976: 132).
921
922
923
924
925
Maister (1976: 129).
926
Maister (1976: 130).
174
Appendix B
·√ ·√ ·√ ·√ . (B12)
The SRL applies to total inventory, if the assumptions stated above of the SRL both as
applied to regular and safety stock hold. For a critical review of these assumptions please
refer to Maister (1976: 125, 132f.), Das (1978: 331f.), McKinnon (1989: 102ff.), and Evers
(1995: 2, 14f.).
It should be checked whether the conditions or assumptions necessary for the SRL to hold
(necessary conditions) are also sufficient conditions, i. e. whether, if the SRL holds, these condi-
tions also automatically apply. This analysis, however, is beyond the scope of this thesis.
Ballou (2004b: 380) incorrectly states “The square-root rule […] measures only the
regular stock reduction, not both regular and safety stock effects […]. Assuming that an
inventory control policy based on the EOQ formula is being followed and that all stocking
points carry the same amount of inventory, the square-root rule can be stated as follows:
√ […]
where
AILT = the optimal amount of inventory to stock, if consolidated into one location in dollars,
pounds, cases, or other units
AILi = the amount of inventory in each of n locations in the same units as AILT
n = the number of stocking locations before consolidation”.
Traditionally the SRL has been mainly applied to safety stock, but as proven above it is also
valid for regular stock or both regular and safety stock, if the respective necessary assumptions
stated above hold. If it is only applied to regular stock as in Ballou's case, all the assumptions of
the SRL as applied to regular stock enumerated above must apply, not only the two mentioned
by Ballou here. Ballou's formulation of the second assumption may confuse, as Maister (1976:
125) assumes “the demand rates at all the field locations are equal”.928
927
Ballou (2004b: 380). Total inventory might also include pipeline, speculative, and obsolete stock (Ballou
2004b: 330f.), which is neglected here.
928
Professor emeritus Ronald H. Ballou, Weatherhead School of Management, Case Western Reserve Uni-
versity, agreed with us in a personal correspondence on May 15, 2007.
175
Appendix C: What Causes the Savings in Regular Stock
through Centralization Measured by the SRL?
The total EOQ in the decentralized system is
∑ . (C1)
The EOQ in the centralized system is
∑ . (C2)
Due to the subadditive property of the square root, the sum of the square root of the de-
mands at the locations i is greater than or equal to the square root of the sum of these de-
mands:
∑ ∑ . (C3)
Therefore, the total EOQ in the decentralized system (the sum of the EOQs of the loca-
tions i) is greater than or equal to the EOQ in the centralized system:
∑ ∑ . (C4)
However, the total EOQ in the decentralized and centralized system are only equal, if
demands or demand forecasts at at least n-1 locations are zero929, which is unusual:
∑ ∑ ∑ ∑
(C5)
∑ 2∑ ∑ ∑ ∑ ∑ 0.
If n-1 di equal zero, (C4) becomes
(C6)
for this single non-zero di. If all di are zero, (C4) becomes
0 . (C7)
Therefore, if demands (at at least two locations) are larger than zero, the EOQ and
therefore the average regular inventory and thus the inventory holding costs are lower in
the centralized system. However, under the above condition the EOQ per decentralized
location i is less than the one for the centralized one, as n > 1:
929
This theoretically questions Wanke and Saliby's (2009: 680) general statement that “[t]he cycle stock
savings are based on the fact that the square-root of the aggregate demand is always smaller than the sum
of the square roots of individual demands”.
176
Appendix C
∑ ∑
∑ . (C8)
The total number of orders in the decentralized system is

∑ ∑ ∑ ∑ . (C9)
The number of orders in the centralized system is

∑ ∑
∑ . (C10)
∑
Due to the subadditive property of the square root, the total number (the sum of the
numbers) of replenishment orders in the decentralized system is greater than or equal to the
number of orders in the centralized one:
∑ ∑ ∑ ∑ . (C11)
In analogy to (C5)-(C7), if demands (at at least two locations) are larger than zero930,
the total number of replenishment orders in the decentralized system is greater than the
number of orders in the centralized one resulting in lower total order fixed costs in the cen-
tralized system. However, in this case the number of orders (lead times) per decentralized
location i is smaller than the number of orders for the centralized one931, as n > 1:
∑ ∑ ∑
∑
∑
(C12)
∑ ∑ .
The savings in regular stock stem from the assumption of an EOQ order policy, constant
fixed costs per order, holding costs (inventory carrying cost rate times product value per
unit), and total demand for all locations before and after centralization. In the centralized
system, usually the total order fixed cost is lower because of less orders and the inventory
holding cost is lower due to a smaller EOQ than in the decentralized system.932
930
Ronen (1990), Zinn et al. (1990), and Evers (1995) neglect this.
931
Cf. Ronen (1990: 132). Zinn et al. (1990: 139) state that “inventory centralization will increase the num-
ber of lead times per year if an EOQ ordering quantity is used”. “[A]n increase in the number of lead
times with the same expected annual demand decreases the average base stock” (Zinn et al. 1990: 141).
This is true for the centralized warehouse compared to each decentralized one.
932
Evers (1995: 5).
177
Appendix D: Tables
Table D.1: Fulfillment of the SRL's Assumptions by Eleven Surveyed Companies
Survey respondents Number 1 2 3 4 5 6 7 8 9 10 11
Country Denmark Germany Germany France Sweden U.S. U.S. Italy U.S. U.S. Germany
Industry Paper Paper Paper Reusable Paper Semicon‐ Wireless Paper
Automotive Utilities Electronics
wholesale production production packaging wholesale ductors telecom wholesale
Employees 65 40,000 200 10,000 60,000 1,850 60,000 9 4,500 6,000 1,652
Position Supply
Vice‐
Director of Functional Chain Supply
Logistics Logistics Marketing Materials Adminis‐ president Director of
Logistics & Integrator Optimiza‐ Chain
Manager Analyst Manager Manager trator Supply Logistics
IT Lead tion Manager
Chain
Manager
Central
2 0 5 10 5 2 0 1 1 2 0
warehouses
Regional
0 100 0 10 66 8 24 2 5 12 21
warehouses
Assumptions of the SRL when applied to
Safety Stock Demands iid
(are indepen‐
dent and
identically 0 1 0 0 0 0 0 0 1 1 1
distributed
random
variables)
Locations'
demands 0 1 1 1 0 1 1 0 1 1 0
uncorrelated
Locations'
demand
0 0 1 0 1 1 0 1 0 1 0
variance the
same
Zero supplier
lead time 0 0 0 0 0 0 0 0 0 0 0
variance
Locations face
same average 0 0 0 1 0 0 0 1 1 1 1
lead time
Lead times iid 1 1 0 0 1 1 1 1 1 1 1
Lead times
and demands
0 0 0 0 0 0 0 0 0 0 0
independent
of each other
Safety factor
approach 0 0 0 0 0 0 1 0 0 0 0
used
Locations use
same safety 1 0 0 0 0 1 0 0 0 0 0
factor
No transship‐
ments
0 0 0 0 0 0 0 1 0 1 0
between
locations
Safety Stock/ Average total
Regular Stock system
demand
0 1 1 0 0 1 0 1 1 1 0
remains the
same after
consolidation
Regular Stock Locations'
demands the 0 0 1 0 0 0 0 1 0 0 0
same
EOQ Order
0 0 1 0 1 1 0 0 1 0 0
Policy
Locations'
fixed order 1 0 1 0 0 1 0 0 1 0 0
cost the same
Locations' per
unit holding 1 0 1 0 0 0 1 0 1 1 0
cost the same
1=Yes, 0=No
178
Appendix D
Table D.2: Comparison of Important Inventory Consolidation Effect, Portfolio Effect, and
Square Root Law Models
179
Appendix D
Continuation of Table D.2
180
Pioneers
Critics
Variable lead time
Transshipments
Optimal consolidation
Network design
!
1
!! "#$%&'(!)*+,-.!$%!&/'!0$(%&!&1!2(13'!&/'!%45#('!(11&!6#7!)89:.!01(!%#0'&;!%&1<=!)88.!#>?!<;<6'!%&1<=!)@8.!5>?'(!<'(&#$>!#%%5A2&$1>%B!C5&!7$&/!0#56&;!#6D'C(#!#>?!21%%$C6'!A$%2($>&%E!
2
!! F/'!89:!G$%!#>!'H<'2&$1>!(#&/'(!&/#>!#!6#7!IJKE!L!&1B!<'>&(#6$M'?!%;%&'A!$%!IJK!&/'!C'%&!25('6;!0(1A!&/'!21$>&!10!3$'7!10!$>3'>&1(;!<1%&%N!)O#%!*+,PQ!RRS.E!
3
!! TH2'<&'?!/16?$>D!#>?!2'>#6&;!<1%&%!#('!%A#66'(!#0&'(!<'>&(#6$M#&$1>E!F/'%'!%#3$>D%!?'<('#%'!7$&/!$><('#%$>D!?'A#>?!<1(('6#&$1>E!U0!?'A#>?%!#('!$?'>&$<#6!#>?!5><1(('6#&'?B!1>'!#(($3'%!#&!&/'!
89:!)T22'>!*+,+.E!
4
!! F/'!21(&016$1!'00'<&!)VT.!?'2'>?%!1>!&/'!%#6'%!<1(('6#&$1>!#>?!A#D>$&5?'!)&/'!('6#&$3'!%$M'%!10!&/'!%&#>?#(?!?'3$#&$1>%!10!?'A#>?.E!F/'!89:!$%!#!%2'<$#6!<#%'!10!&/'!VTB!$0!&/'!A#D>$&5?'!$%!
'45#6!&1!*!#>?!&/'!%#6'%!<1(('6#&$1>!$%!'45#6!&1!W!)X$>>!'&!#6E!*+P+.E!
5
!! O'0$>$>D!%'(3$<'!6'3'6!#%!G&/'!?'%$('?!0(#<&$1>!10!?'A#>?!%5226$'?!0(1A!%&1<=N!G>1!D'>'(#6!('%56&%!<#>!C'!?'($3'?!<1><'(>$>D!&/'!?'%$(#C$6$&;!10!$>3'>&1(;!<'>&(#6$M#&$1>Y?'<'>&(#6$M#&$1>!
7$&/15&!=>17$>D!&/'!1(?'(!45#>&$&$'%!#>?!&/'!2(1C#C$6$&;!?$%&($C5&$1>%!$>3163'?N!)91>'>!*++WQ!*RZ.E!
6
!! "#/A15?!)*++[.!?'%<($C'%!&/'!V1(&016$1!\5#>&$&;!T00'<&!)V\T.B!&/'!V1(&016$1!@1%&!T00'<&!)V@T.B!#>?!#>!12&$A#6!<1>%16$?#&$1>!%</'A'E!852'(]<1>%16$?#&$1>!$%!>1&!#67#;%!12&$A#6E!
7
!! T3'(%!#>?!^'$'(!)*++R.!<1>%$?'(!6'#?!&$A'!5><'(&#$>&;B!5>'45#6!#3'(#D'!6'#?!&$A'%B!#>?!<'>&(#6$M#&$1>!&1!%'3'(#6!61<#&$1>%!)A56&$26'!<1>%16$?#&$1>!21$>&%.E!F/'!'45#&$1>!1>!2#D'!**S!$%!$>]
<1(('<&Q!U&!%/156?!('#?Q!=!_!%4(&)*Y)*]C'&#..!)2'(%1>#6!<1(('%21>?'><'!7$&/!V(10'%%1(!T3'(%!1>!`56;!-B![WW+.E!
8
!! F#661>!)*++R.!<1>%$?'(%!3#($#C6'!?'A#>?!#>?!('26'>$%/A'>&!6'#?!&$A'%!#>?!<1(('6#&$1>!10!?'A#>?!?5($>D!6'#?!&$A'E!88!%#3$>D%!0(1A!<'>&(#6$M#&$1>!?'2'>?!1>!&/'!<1(('6#&$1>!10!%#6'%!#>?!
&/'!(#&$1!10!&/'!%&#>?#(?!?'3$#&$1>%!10!?'A#>?!?5($>D!6'#?!&$A'!)A#D>$&5?'.E!
9
!! T3'(%!)*++S.!<1>%$?'(%!&/'!@1>%16$?#&$1>!T00'<&!)@T.!5>?'(!3#($15%!#%%5A2&$1>%E!F/'!@T!%/17%!&/'!$A2#<&!10!$>3'>&1(;!2116$>D!1>!C1&/!88!#>?!@8!)T3'(%!*++SQ!*.E!@8!?'&'(A$>'?!C;!&/'!
Ta\!#22(1#</!<#>!C'!?'<('#%'?!&/(15D/!<'>&(#6$M#&$1>B!C5&!2($A#(;!%#3$>D%!('%56&!0(1A!('?5<&$1>%!$>!88E!b'>'(#66;B!&/'('!#('!>1!('?5<&$1>%!$>!@8B!$0!&/'!('26#<'A'>&!2($><$26'!$%!016617'?E!
F/'!#%%5A2&$1>%!#('!>1&!6$='6;!&1!C'!015>?!$>!2(#<&$<'!<166'<&$3'6;E!U>3'>&1(;!$><('#%'%!7$&/!&/'!>5AC'(!10!61<#&$1>%E!
10
!! @0E!"#$%&'(!)*+,-.E!
11
!! F/'!6#%&!<1><65?$>D!C(#<='&!$>!&/'!>5A'(#&1(!$%!A$%%$>D!$>!T3'(%!#>?!^'$'(!)*++RQ!**Z.E!
12
!! c$&/!'45#6!2#(&$&$1>$>D!10!?'A#>?!$>!#??$&$1>!&1!"#$%&'(d%!)*+,-.!#%%5A2&$1>%B!T3'(%!#>?!^'$'(!)*++R.!#(($3'!#&!#!('3$%'?!89:Q!F/'('!#('!>1!05(&/'(!%#3$>D%!$>!88!#0&'(!&/'!0$(%&!<1>%16$?#]
&$1>E!F/'!('?5<&$1>!$>!88!?'2'>?%!1>6;!1>!&/'!>5AC'(!10!?'<'>&(#6$M'?!0#<$6$&$'%E!
13
!! T22'>!)*+,+.!<1>%$?'(%!&/'!'>?]10]2'($1?!$>3'>&1(;!C'01('!#!>'7!1(?'(!#(($3'%B!7/$</!A#;!C'!%''>!#%!&/'!%#0'&;!%&1<=!)8$63'(!'&!#6E!*++PQ![RZ.B!#>?!1>6;!G%&#&$%&$<#6!'<1>1A$'%!10!%<#6'NB!>1&!
&/'!1(?'(!45#>&$&;!'00'<&E!e16?$>D!#>?!%/1(&#D'!<1%&%!#('!2(121(&$1>#6!&1!&/'!%&#>?#(?!?'3$#&$1>!10!?'A#>?E!
14
!! T3'(%!)*++SQ!R.E!
15
!! U>!T22'>!)*+,+.B!#&!&/'!C'D$>>$>D!10!'3'(;!2'($1?!61<#&$1>!$!(#$%'%!$&%!$>3'>&1(;!&1!&/'!6'3'6!A$>$A$M$>D!$&%!'H2'<&'?!/16?$>D!#>?!%/1(&#D'!<1%&%!01(!&/'!('%2'<&$3'!2'($1?B!7/$</!$%!&/'!<1(]
('%21>?$>D!>'7%3'>?1(!2(1C6'Ad%!12&$A#6!%165&$1>E!F/$%!$%!<#66'?!C#%']%&1<=B!1(?'(]52]&1]6'3'6B!1(!)9B!8.!216$<;E!
16
!! GU0!&/'!6'3'6!10!$>3'>&1(;!1>!/#>?!265%!1>!1(?'(!$%!6'%%!&/#>!&/'!('1(?'(!21$>&B!#!('1(?'(!$%!26#<'?!7$&/!#>!15&%$?'!%15(<'!#>?!&/'!1>]1(?'(!$>3'>&1(;!6'3'6!$%!$><('#%'?N!)T3'(%!*++-Q!**,.E!
17
!! 91>'>!)*++WQ!*RW.!<1>%$?'(%!#>!Ga(?'(!\5#>&$&;Y9'1(?'(!V1$>&!)\B9.!$>3'>&1(;!<1>&(16!216$<;E!O'A#>?!?5($>D!%&1<=!('26'>$%/A'>&!6'#?!&$A'!/#%!%1A'!%&#&$%&$<#6!?$%&($C5&$1>!?5'!&1!3#($#]
C$6$&;!10!?'A#>?B!6'#?!&$A'B!1(!C1&/E!F/'!1(?'(!45#>&$&;!$%!?'&'(A$>'?!C;!&/'!T<1>1A$<!a(?'(!\5#>&$&;!)Ta\.!#>?!&/'!('1(?'(!21$>&B!9B!<1>%$%&%!10!#3'(#D'!?'A#>?!?5($>D!6'#?!&$A'!265%!#!
%#0'&;!%&1<=B!7/$</!$%!#!<'(&#$>!>5AC'(!)=.!10!%&#>?#(?!?'3$#&$1>%!10!?'A#>?!?5($>D!6'#?!&$A'NE!85</!#>!1(?'(!216$<;!$%!#6%1!<#66'?!)%B!4.E!
18
!! L66!61<#&$1>%!<1>&(16!$>3'>&1($'%!C;!&/'!c$6%1>!:1&!8$M'!f1(A56#!1(!Ta\!C'01('!#>?!#0&'(!2116$>D!)"#$%&'(!*+,-Q!*R*.E!
19
!! F/'!1(?'(!45#>&$&;!$%!'45#6!&1!&/'!#3'(#D'!?'A#>?!?5($>D!6'#?!&$A'!)T3'(%!*++SQ!*Z.E!F/'!('26#<'A'>&!2($><$26'!1(!%&1<=]&1]?'A#>?!216$<$'%!<#>!C'!%5C%5A'?!5>?'(!1(?'(]52]&1]6'3'6!A1?]
'6%!)c#>='!#>?!8#6$C;![WW+Q!-PW.E!
20
!! X$>>!'&!#6E!)*++W.E!
21
!! T22'>!)*+,+.!<1>%$?'(%!#!%$>D6']2'($1?!%$>D6'!2(1?5<&!A56&$]61<#&$1>!>'7%C1;!2(1C6'AE!
22
!! L66!61<#&$1>%!5&$6$M'!&/'!%#0'&;]0#<&1(!)=ı.!#22(1#</!&1!%'&!%#0'&;!%&1<=!C'01('!#>?!#0&'(!2116$>D!)"#$%&'(!*+,-Q!*[+.E!
23
!! L66!61<#&$1>%!5%'!&/'!%#A'!%#0'&;!%&1<=!A56&$26'!=!C'01('!#>?!#0&'(!2116$>D!)"#$%&'(!*+,-Q!*R[.E!
24
!! L66!61<#&$1>%!100'(!&/'!%#A'!%'(3$<'!6'3'6!)91>'>!*++W.E!
25
!! X$>>!'&!#6E!)*+P+.!5%'!&/'!2(1C#C$6$&;!10!>1&!(5>>$>D!15&!10!%&1<=!?5($>D!6'#?!&$A'!#%!&/'!<5%&1A'(!%'(3$<'!6'3'6!A'#%5('E!
26
Appendix D
181
!! 91>'>!)*++W.!5%'%!&/'!0(#<&$1>!10!?'A#>?!%5226$'?!0(1A!%&1<=!?5($>D!6'#?!&$A'!#%!&/'!<5%&1A'(!%'(3$<'!6'3'6!A'#%5('E!
Appendix D
182
Appendix D
183
Appendix D
Table D.2 shows the different SRL and PE models and the assumptions they are based
upon. With the help of this table one can choose the model best fit for assessing inventory
changes from centralization or decentralization for given conditions. The corresponding
formulae for calculating the inventory savings achievable by inventory pooling can be
found on the respective publication's page noted in the first or second row of this table.
In order to facilitate readability the assumptions are grouped according to the rubrics
stock-, inventory-policy-, demand-, and lead-time-related. The columns or models are
arranged chronologically according to their year of publication, so that the development
and adding or dropping of assumptions over time can be retraced where possible.
The models are also grouped by colors according to their main contribution or focus,
although this is not too definite, as they are very different or overlapping and the terms
SRL and PE are sometimes used interchangeably.
Maister (1976) showed the SRL for the EOQ and the safety factor approach for both
cycle and safety stock, Eppen (1979) for the newsvendor model and expected costs, Zinn et
al. (1989) the PE as a generalization of the SRL, Mahmoud (1992) the Portfolio Quantity
and Portfolio Cost Effect, Evers (1995) the Consolidation Effect, Caron and Marchet
(1996) the impact of centralization on safety stock in a special two-echelon system, i. e. the
ratio of safety stock levels in a coupled system and an independent system, first (pioneers).
Das (1978) and Ronen (1990)933 (critics) critically review Maister's (1976) SRL and Zinn
et al.'s (1989) PE respectively. Evers and Beier (1993, 1998) and Tallon (1993) consider
variable lead times, Evers (1996, 1997) transshipments, Tyagi and Das (1998), Das and
Tyagi (1999), Wanke (2009), and Wanke and Saliby (2009) optimal consolidation
schemes, and Croxton and Zinn (2005) inventory pooling effects with the SRL in network
design.
Of course all researchers may be considered critical pioneers as they all contribute
something novel and critically review previous work, which is the essence of scientific
research. Mahmoud (1992: 198, 212) also considers the optimal consolidation scheme and
criticizes that the PE is not sufficient to define it. Evers and Beier (1993) and Evers (1995)
criticize Maister's (1976) SRL. Evers and Beier (1993) and Tyagi and Das (1998) consider
multiple consolidation points and the equal partitioning of demand assumption. Wanke and
Saliby (2009) also deal with regular transshipments. Important distinguishing assumptions
made or not made by the various models are highlighted in grey.
933
Ronen (1990) and Zinn et al. (1989, 1990) have a scientific dispute. For details please refer to the respec-
tive publications.
184
Appendix D
Table D.3: Conditions Favoring the Various Risk Pooling Methods
185
Appendix D
186
Appendix D
187
Appendix D
188
Appendix D
189
Appendix D
190
Appendix D
191
Appendix D
192
Appendix D
193
Appendix D
194
Appendix D
Table D.4 The Risk Pooling Methods' Advantages, Disadvantages, Performance, and Trade-
Offs
195
Appendix D
196
Appendix D
197
Appendix D
198
Appendix D
199
Appendix D
200
Appendix D
201
Appendix E: Subadditivity Proof for the Order Quantity
Function (5.1)
The total system-wide order quantity in the centralized system is no greater than the one in
the decentralized system, because the order quantity function (5.1) is subadditive:
∑
, ∑ , . (5.4)
In order to prove this, we first resort to two warehouses with forecast demand minus
available inventory (net demand) and and seek to prove:
, , , . (E1)
The order quantity function (5.1) is a compound function consisting of an outer

maximum function and an inner ceiling function, which we henceforth denote by and ,
respectively. The ceiling expression is a constant.
Recall that a function is called subadditive, if it obeys the inequality

(E2)
for any , from the function's domain.
We first show that is subadditive,
i. e. that the following holds for any , :
. (E3)
This can be verified by a case differentiation:
For s d , s d , (E4)
s d , s d
for s d ,s d , (E5)
s d ,s d
where notation a b stands for „a divides b“ or, equivalently, “b is a multiple of a” and
designates negation.
We now prove that , is subadditive as well, i. e. that the following
holds for any , :
, , , . (E6)
g ,g
For g ,g c , , , . (E7)
g c, g
For g c, g c , , , . (E8)
202
Appendix E
Then for the compound function holds:

, (E9)
since
(E10)
,
where the first inequality is due to subadditivity of and non-decreasing behavior of and
the second inequality is due to subadditivity of .
It is now straightforward to generalize (E9) to any number of summands. Given
, we by induction on n have:
(E11)
,
which hence gives
∑ ∑ (E12)
and proves (5.4) to hold.
203
Bibliography
Abramowitz, M. & Stegun, I. A., eds. (1972), Handbook of Mathematical Functions with
Formulas, Graphs, and Mathematical Tables, Dover.
Ackerman, K. B. & Brewer, A. M. (2001), Warehousing: A key link in the supply chain, in
Brewer, A. M., ed., Handbook of Logistics and Supply Chain Management, Elsevier
Science, pp. 225-236.
Adam, D. (1979), Kostendegressionen und –progressionen, in Kern, W., ed., HWB der
Produktionswirtschaft, Poeschel, pp. 939-955.
Adam, E. E. & Ebert, R. J. (1976), 'A comparison of human and statistical forecasting',
AIIE Transactions 8, 120-127.
AFSAC (2001), U.S. DEPARTMENT OF AIR FORCE, 212-448-0496, AGENCY
GROUP 09, REGION GROUP 06, 'Virtual warehouse offers global parts shopping',
FDCH Regulatory Intelligence Database, 03/14/2001, http://search.ebscohost.com/
login.aspx?direct=true&db=buh&AN=32W20019200005928&site=ehost-live, accessed
03/17/2008.
Alderson, W. (1957), Marketing Behaviour and Executive Action, Richard D. Irwin.
Alfaro, J. A. & Corbett, C. J. (2003), 'The value of SKU rationalization in practice (the
pooling effect under suboptimal inventory policies and nonnormal demand)',
Production & Operations Management 12(1), 12-29.
Alfredsson, P. & Verrijdt, J. (1999), 'Modeling emergency supply flexibility in a two-
echelon inventory system', Management Science 45(10), 1416-1431.
Alicke, K. (2005), Planung und Betrieb von Logistiknetzwerken: Unternehmensübergrei-
fendes Supply Chain Management, Springer.
Alisch, K.; Arentzen, U. & Winter, E., eds. (2004), Gabler Wirtschaftslexikon, Gabler.
Alvarez, A. (2007), 'Competing with the internet', JCK 178(10), 117.
Anand, K. S. & Girotra, K. (2007), 'The strategic perils of delayed differentiation',
Management Science 53(5), 697-712.
Andres, H. P. & Zmud, R. W. (2001), 'A contingency approach to software project
coordination', Journal of Management Information Systems 18(3), 41-70.
Anupindi, R. & Bassok, Y. (1999), 'Centralization of stocks: Retailers vs. manufacturer',
Anupindi, R. & Jiang, L. (2008), 'Capacity investment under postponement strategies,
market competition, and demand uncertainty', Management Science 54(11), 1876-1890.
Anupindi, R.; Bassok, Y. & Zemel, E. (2001), 'A general framework for the study of
decentralized distribution systems', Manufacturing & Service Operations Management
3(4), 349-369.
Anupindi, R.; Chopra, S.; Deshmukh, S. D.; Van Mieghem, J. A. & Zemel, E. (2006),
Managing Business Process Flows: Principles of Operations Management, Pearson/
Prentice Hall.
204
Bibliography
Anupindi, R.; Dada, M. & Gupta, S. (1998), 'Estimation of consumer demand with stock-
out based substitution: An application to vending machine products', Marketing Science
17(4), 406-423.
Archibald, T. W. (2007), 'Modelling replenishment and transshipment decisions in periodic
review multilocation inventory systems', Journal of the Operational Research Society
58(7), 948-956.
Archibald, T. W.; Black, D. & Glazebrook, K. D. (2009), 'An index heuristic for
transshipment decisions in multi-location inventory systems based on a pairwise
decomposition', European Journal of Operational Research 192(1), 69-78.
Archibald, T. W.; Sassen, S. A. E. & Thomas, L. C. (1997), 'An optimal policy for a two
depot inventory problem with stock transfer', Management Science 43(2), 173-183.
Arnold, J. & Köchel, P. (1996), Evolutionary optimization of a multi-location inventory
model with lateral transshipments, in Grubbström, R. W., ed., 'Proceedings of the 9th
International Working Seminar on Production Economics', Vol. 2, University of
Innsbruck, pp. 401-412.
Arnold, J.; Köchel, P. & Uhlig, H. (1997), With parallel evolution towards the optimal
order policy of a multi-location inventory with lateral transshipments, in Papachristos,
S. & Ganas, I., eds., 'Proceedings of the 3rd ISIR Summer School on Inventory
Modelling in Production and Supply Chains', International Society for Intentory
Research, pp. 1-14.
Ashayeri, J. & Selen, W. (2005), 'An application of a unified capacity planning system',
International Journal of Operations & Production Management 25(9), 917-937.
Ashford, M. (1997), Cases 4 and 5 - Hasbro Europe and Nike: Developing European
logistics strategy, in Taylor, D. H., ed., Global Cases in Logistics and Supply Chain
Management: Teachers' Manual, International Thomson Business Press, pp. 16-23.
ATL (2009), 'NCloc Deutschland- und europaweite Standortoptimierung', Fraunhofer
Institut Integrierte Schaltungen, Arbeitsgruppe für Technologien der Logistik-
Dienstleistungswirtschaft ATL, Nuremberg, http://www.atl.fhg.de/fileadmin/data/pro
dukte_und_dienstleistungen/eus/NCloc_Flyer.pdf, accessed 07/18/2009.
Aviation Week & Space Technology (2006), 'A380 support', Aviation Week & Space
Technology 165(12), 16.
Aviv, Y. & Federgruen, A. (1997), 'Stochastic inventory models with limited production
capacity and periodically varying parameters', Probability in the Engineering and
Informational Sciences 11(1), 107-135.
Aviv, Y. & Federgruen, A. (2001a), 'Capacitated multi-item inventory systems with
random and seasonally fluctuating demands: Implications for postponement strategies',
Aviv, Y. & Federgruen, A. (2001b), 'Design for postponement: A comprehensive
characterization of its benefits under unknown demand distributions', Operations
Research 49(4), 578-598.
Axsäter, S. (1990), 'Modelling emergency lateral transshipments in inventory systems',
Axsäter, S. (2003a), 'A new decision rule for lateral transshipments in inventory systems',
205
Bibliography
Axsäter, S. (2003b), 'Evaluation of unidirectional lateral transshipments and substitutions

in inventory systems', European Journal of Operational Research 149(2), 438-447.
Badinelli, R. & Schwarz, L. (1988), 'Backorders optimization in a one-warehouse n-
identical retailer distribution system', Naval Research Logistics Quarterly 35(5), 427-
440.
Bagchi, P. K. & Skjoett-Larsen, T. (2005), 'Supply chain integration: A European survey',
International Journal of Logistics Management 16(2), 275-294.
Bagchi, U. & Gutierrez, G. (1992), 'Effect of increasing component commonality on
service level and holding cost', Naval Research Logistics 39(6), 815-832.
Bailey, J. & Rabinovich, E. (2005), 'Internet book retailing and supply chain management:
An analytical study of inventory location speculation and postponement', Logistics and
Transportation Review 41(3), 159-177.
Baker, K. R. (1985), 'Safety stock and component commonality', Journal of Operations
Management 6(1), 13-22.
Baker, K. R.; Magazine, M. J. & Nuttle, H. L. W. (1986), 'The effect of commonality on
safety stock in a simple inventory model', Management Science 32(8), 982-988.
Ballou, R. H. & Burnetas, A. (2000), 'Planning virtual inventories', Technical
Memorandum 742, Department of Operations, Weatherhead School of Management,
Case Western Reserve University, Cleveland, OH.
Ballou, R. H. & Burnetas, A. (2003), 'Planning multiple location inventories', Journal of
Business Logistics 24(2), 65-89.
Ballou, R. H. (1979), 'An extended distribution analysis to support marketing and
production planning', Journal of Business Logistics 1(1), 63-75.
Ballou, R. H. (1981), 'Estimating and auditing aggregate inventory levels at multiple
stocking points', Journal of Operations Management 1(3), 143-153.
Ballou, R. H. (2000), 'Evaluating inventory management performance using a turnover
curve', International Journal of Physical Distribution & Logistics Management 30(1),
72-88.
Ballou, R. H. (2003a), 'PowerPoint presentation chapter 8: Forecasting supply chain
requirements', 27/08/2003, http://www.pearsonhighered.com/educator/product/Instruc
tors-Download-Site-with-Instructors-Manual-5E/9780131428010.page, accessed 23/07/
2009.
Ballou, R. H. (2003b), 'PowerPoint presentation chapter 9: Inventory policy decisions',
27/08/2003, http://www.pearsonhighered.com/educator/product/Instructors-Download-
Site-with-Instructors-Manual-5E/9780131428010.page, accessed 02/12/2009.
Ballou, R. H. (2004a), 'Expressing inventory control policy in the turnover curve',
Technical Memorandum 794, Department of Operations, Weatherhead School of
Management, Case Western Reserve University, Cleveland, OH.
Ballou, R. H. (2004b), Business Logistics / Supply Chain Management, Pearson/Prentice
Hall.
Ballou, R. H. (2004c), 'Logware: Selected Computer Programs for Logistics/Supply Chain
Planning: Version 5.0', Weatherhead School of Management, Case Western Reserve
University, Cleveland, OH.
206
Bibliography
Ballou, R. H. (2005), 'Expressing inventory control policy in the turnover curve', Journal
of Business Logistics 26(2), 143-164.
Bandy, D. B. (2005), 'In-class simulation of pooling safety stock', Decision Sciences
Journal of Innovative Education 3(2), 375-380.
Bartholdi, J. J. & Kemahlioğlu-Ziya, E. (2005), Using Shapley value to allocate savings in
a supply chain, in Geunes, J. & Pardalos, P. M., eds., Supply Chain Optimization,
Springer, pp. 169-208.
Baruch, Y. (1999), 'Response rate in academic studies: A comparative analysis', Human
Relations 42(4), 421-438.
Bassok, Y.; Anupindi, R. & Akella, R. (1999), 'Single-period multiproduct inventory
models with substitution', Operations Research 47(4), 632-642.
Battezzati, L. & Magnani, R. (2000), 'Supply chains for FMCG and industrial products in
Italy: Practices and the advantage of postponement', International Journal of Physical
Distribution & Logistics Management 30(5), 413-424.
Bauer, H. (1974), Wahrscheinlichkeitstheorie und Grundzüge der Maßtheorie, de Gruyter.
Bender, T.; Hennes, H.; Kalcsics, J.; Melo, T. M. & Nickel, S. (2004), 8 Location software
and interface with GIS and supply chain management, in Drezner, Z. & Hamacher, H.
W., eds., Facility Location: Applications and Theory, Springer, pp. 233-274.
Bendoly, E. (2004), 'Integrated inventory pooling for firms servicing both on-line and store
demand', Computers & Operations Research 31(9), 1465-1480.
Bendoly, E.; Blocher, D.; Bretthauer, K. M. & Venkataramanan, M. (2007), 'Service and
cost benefits through clicks-and-mortar integration: Implications for the
centralization/decentralization debate', European Journal of Operational Research
180(1), 426-442.
Benjaafar, S. & Kim, J.-S. (2001), 'When does higher demand variability lead to lower
safety stocks?', technical report, Department of Mechanical Engineering, University of
Minnesota, Minneapolis, MN.
Benjaafar, S.; Cooper, W. L. & Kim, J.-S. (2005), 'On the benefits of pooling in
production-inventory systems', Management Science 51(4), 548-565.
Benjaafar, S.; ElHafsi, M. & De Véricourt, F. (2004a), 'Demand allocation in multiple-
product, multiple-facility, make-to-stock systems', Management Science 50(10), 1431-
1448.
Benjaafar, S.; Kim, J.-S. & Vishwanadham, N. (2004b), 'On the effect of product variety in
production-inventory systems', Annals of Operations Research 126(1-4), 71-101.
Benton, W. C. & Krajewski, L. (1990), 'Vendor performance and alternative
manufacturing environments', Decision Sciences 21(2), 403-415.
Ben-Zvi, N. & Gerchak, Y. (2007), 'Inventory centralization when shortage costs differ:
Priorities and costs allocation', technical report, Department of Industrial Engineering,
Tel-Aviv University, Tel-Aviv, Israel.
Bernstein, F.; Kök, A. G. & Xie, L. (2009), 'The role of component commonality in
product assortment decisions', technical report, Fuqua School of Business, Duke
University, Durham, NC, 08/06/2009, http://faculty.fuqua.duke.edu/~agkok/papers/Com
monality.pdf, accessed 01/03/2010.
207
Bibliography
Berry, W. L.; Tallon, W. J. & Boe, W. J. (1992), 'Product structure analysis for the master
scheduling of assemble-to-order products', International Journal of Operations &
Production Management 12(11), 24-41.
Bertrand, L. P. & Bookbinder, J. H. (1998), 'Stock redistribution in two-echelon logistics
systems', Journal of the Operational Research Society 49(9), 966-975.
Beuthe, M. & Kreutzberger, E. (2001), Consolidation and trans-shipment, in Brewer, A.
M.; Button, K. J. & Hensher, D. A., eds., Handbook of Logistics and Supply Chain
Management, Elsevier Science, pp. 239-252.
Bhaumik, P. K. & Kataria, S. (2006), 'Lateral transshipment for managing excesses and
shortages in a multilocation inventory system: A case study of Timex Watches Ltd.',
International Journal of Services Technology & Management 7(5/6), 602-614.
Bickart, B. & Schmittlein, D. (1999), 'The distribution of survey contact and participation
in the United States: Constructing a survey-based estimate', Journal of Marketing
Research 36(2), 186-294.
Bidgoli, H., ed. (2010), The Handbook of Technology Management: Supply Chain
Management, Marketing and Advertising, and Global Management, Wiley.
Blaikie, N. W. H. (1991), 'A critique of the use of triangulation in social research', Quality
& Quantity 25(2), 115-136.
Böhnlein, C.-B. & Lünemann, P. (2008), Supply Chain Management in der Praxis - Status
2007, Vol. 8, Oxygon.
Bohr, K. (1996), Economies of scale and economies of scope, in Kern, W. & Schröder, H.-
H., eds., Handwörterbuch der Produktionswirtschaft, Schäffer-Poeschel, pp. 375-386.
Bollapragada, S.; Akella, R. & Srinivasan, R. (1998), 'Centralized ordering and allocation
policies in a two-echelon system with non-identical warehouses', European Journal of
Operational Research 106(1), 74-81.
Boone, C. A.; Craighead, C. W. & Hanna, J. B. (2007), 'Postponement: An evolving
supply chain concept', International Journal of Physical Distribution & Logistics
Management 37(8), 594-611.
Bowersox, D. J. & Closs, D. J. (1996), Logistical Management: The Integrated Supply
Chain Process, McGraw-Hill.
Bowersox, D. J.; Closs, D. J. & Cooper, M. B. (2002), Supply Chain Logistics
Management, McGraw-Hill/Irwin.
Bowersox, D. J.; Closs, D. J. & Helferich, O. K. (1986), Logistical Management: A
Systems Integration of Physical Distribution, Manufacturing Support, and Materials
Procurement, MacMillan.
Bowersox, D. J.; Daugherty, P. J.; Dröge, C. L.; Germain, R. N. & Rogers, D. S. (1992),
Logistical Excellence: It's Not Business as Usual, Digital Press.
Bowersox, D. J.; Smykay, E. W. & La Londe, B. J. (1968), Physical Distribution
Management: Logistics Problems in the Firm, MacMillan.
Bowersox, D. J.; Stank, T. P. & Daugherty, P. J. (1999), 'Lean launch: Managing product
introduction risk through response-based logistics', Journal of Product Innovation
Management 16(6), 557-568.
208
Bibliography
Boyabatli, O. & Toktay, L. B. (2004), 'Operational hedging: A review with discussion',

technical report, INSEAD, Fontainebleau, France, January 2004, http://www.prism.ga
tech.edu/~bt71/articles/pompaper.pdf, accessed 05/20/2009.
Boysen, N. & Scholl, A. (2009), 'A general solution framework for component-
commonality problems', Business Research 2(1), 86-106.
Bramel, J. & Simchi-Levi, D. (1997), The Logic of Logistics: Theory, Algorithms, and
Applications for Logistics Management, Springer.
Brandimarte, P. & Zotteri, G. (2007), Introduction to Distribution Logistics, Wiley.
Breakwell, G. M.; Hammond, S.; Fife-Schaw, C. & Smith, J. A. (2007), Research Methods
in Psychology, Sage.
Bretzke, W.-R. (2008), Logistische Netzwerke, Springer.
Bretzke, W.-R. (2010), Logistische Netzwerke, Springer.
Bridgefield Group, Inc. (2006), 'Bridgefield Group ERP/supply chain glossary', http://www
.bridgefieldgroup.com/bridgefieldgroup/glos5.htm, accessed 23/07/2009.
Brown, A. O.; Lee, H. L. & Petrakian, R. (2000), 'Xilinx improves its semiconductor
supply chain using product and process postponement', Interfaces 30(4), 65-80.
Bryman, A. (1992), Quantitative and qualitative research: Further reflection on their
integration, in Brannen, J., ed., Mixing Methods: Qualitative and Quantitative Research,
Avebury, pp. 57-78.
Bucklin, L. P. (1965), 'Postponement, speculation and the structure of distribution
channels', Journal of Marketing Research 2(1), 26-31.
Bunn, D. & Wright, G. (1991), 'Interaction of judgemental and statstical forecasting
methods: Issues & analysis', Management Science 37(5), 501-518.
Burchell, B. (2009), 'Pooling pays', Aviation Week & Space Technology 171(14), 38.
Burer, S. & Dror, M. (2006), 'Newsvendor games: Convex optimization of centralized
inventory operations', 09/28/2006, http://dollar.biz.uiowa.edu/~sburer/papers/014-inven
tory.pdf, accessed 03/28/2008.
Burton, J. & Banerjee, A. (2005), 'Cost-parametric analysis of lateral transshipment
policies in two-echelon supply chains', International Journal of Production Economics
93-94(1), 169-178.
Busse von Colbe, W. & Laßmann, G. (1986), Betriebswirtschaftstheorie, Band 1,
Grundlagen, Produktions- und Kostentheorie, Springer.
BVdDP (2009), Bundesverband des Deutschen Papiergroßhandels e.V., 'Jahresbericht
2008', http://www.verband-papiergrosshandel.de/bvddp-jahresbericht.html, accessed
07/16/2009.
BVdDP (2010a), Bundesverband des Deutschen Papiergroßhandels e.V., 'Zahlen und
Fakten', http://www.verband-papiergrosshandel.de/pdf/BVdDP.pdf, accessed 05/11/
2010.
BVdDP (2010b), Bundesverband des Deutschen Papiergroßhandels e.V., 'Jahresbericht
2009', http://www.verband-papiergrosshandel.de/bvddp-jahresbericht.html, accessed
07/29/2010.
Cachon, G. & Terwiesch, C. (2009), Matching Supply with Demand: An Introduction to
Operations Management, McGraw-Hill.
209
Bibliography
Cachon, G. P. & Lariviere, M. A. (2005), 'Supply chain coordination with revenue-sharing

contracts: Strengths and limitations', Management Science 51(1), 30-44.
Cachon, G. P. (2003), Supply chain coordination with contracts, in de Kok, A. G. &
Graves, S. C., eds., Handbooks in Operations Research and Management Science (Vol.
11): Supply Chain Management: Design, Coordination and Operation, North-Holland,
ch. 6.
Cachon, G. P. (2004), 'Empiricism in supply chain management research', The Wharton
School, University of Pennsylvania, Supply Chain Roundtable, July 2004, http://mba.
tuck.dartmouth.edu/digital/Programs/CorporateEvents/SupplyChainThoughtLeaders/ses
sion1bSlides.pdf, accessed 10/20/2008.
Cachon, G. P.; Randall, T. & Schmidt, G. M. (2007), 'In search of the bullwhip effect',
Manufacturing & Service Operations Management 9(4), 457-479.
Caddy, I. & Helou, M. (2000), Virtual inventories: Impact on inventory and logistics
management, in Dahiya, S. B., ed., The Current State of Business Disciplines, Vol. 4,
Management-I, Rohtak, pp. 1707-1723.
Campbell, J. F. (1993), 'One-to-many distribution with transshipments: An analytic model',
Transportation Science 27(4), 330-340.
Campbell, J. Y.; Lettau, M.; Malkiel, B. G. & Xu, Y. (2001), 'Have individual stocks
become more volatile? An empirical exploration of idiosyncratic risk', Journal of
Finance 56(1), 1-43.
Capgemini U.S. LLC. (2003), 'New supply chain study finds postponement strategies
critical for reducing demand uncertainty and improving customer satisfaction: APICS,
Cap Gemini Ernst & Young, and Oracle study cites perceived technology limitations
and lack of postponement knowledge as top inhibitors', 10/07/2003, http://www.us.cap
gemini.com/news/current_news.asp?ID=335&PRyear=2003, accessed 04/20/2009.
Caron, F. & Marchet, G. (1996), 'The impact of inventory centralization/decentralization
on safety stock for two-echelon systems', Journal of Business Logistics 17(1), 233-257.
Carter, P. L. & Monczka, R. M. (1978), 'MRO inventory pooling', Journal of Purchasing
& Materials Management 14(3), 27-33.
Carver, R. (1978), 'The case against statistical significance testing', Harvard Educational
Review 48, 378-399.
Cattani, K. D. (2000), 'Demand pooling effects of a universal product when demand is
from distinct markets', technical report, The University of North Carolina at Chapel
Hill.
Cattrysse, D.; van Oudheusden, D. & Wong, H. (2009), 'Inventory pooling of repairable
spare parts', Katholieke Universiteit Leuven, Centre for Industrial Management, http://
cib.kuleuven.be/research/reskt12.htm, accessed 03/03/2009.
Caux, C.; David, F. & Pierreval, H. (2006), 'Implementation of delayed differentiation in
batch process industries: A standardization problem', International Journal of
Production Research 44(16), 3243-3255.
Caves, D. W.; Christiansen, L. R. & Tretheway, M. W. (1984), 'Economies of density
versus economies of scale: Why trunk and local service airline costs differ', Rand
Journal of Economics 15(4), 471-488.
210
Bibliography
Cecere, L. & Keltz, H. (2008), 'Is your CP organization ready for rising demand
variability?', 05/12/2008, http://www.terratechnology.com/assets/Uploads/Is-Your-CP-
Organization-Ready-for-Rising-Demand-Variability.pdf, accessed 08/18/2010.
Chain Drug Review (2009a), 'SKU rationalization a new hurdle', Chain Drug Review
31(14), 22.
Chain Drug Review (2009b), 'Innovative items find way amid SKU rationalization', Chain
Drug Review 31(17), 24.
Chandler, A. D. (1990), Scale and Scope - The Dynamics of Industrial Capitalism, Harvard
University Press.
Chandra, C. & Kamrani, A. K. (2004), Mass Customization: A Supply Chain Approach,
Kluwer.
Chang, P. & Lin, C. (1991), 'On the effect of centralization on expected costs in a multi-
location newsboy problem', Journal of the Operational Research Society 42(11), 1025-
1030.
Chang, P.-L.; Lin, C.-T. & Shen, P.-D. (1996), 'A note on centralized effect on expected
costs in a multi-location newsboy problem', International Journal of Information and
Management Sciences 7(1), 25-31.
Chauhan, S. S.; Martel, A. & D'Amour, S. (2008), 'Roll assortment optimization in a paper
mill: An integer programming approach', Computers & Operations Research 35(2),
614-627.
Chen, C. & Ren, Y. (2007), 'Predictive methods for using capacity data to estimate market
shares and the extent of risk pooling by airline alliance partners under parallel
codesharing', Transportation Journal 46(4), 21-35.
Chen, F. C.-Y. & Chen, C. (2003), 'The effects of strategic alliances and risk pooling on
the load factors of international airline operations', Transportation Research: Part E
39(1), 19-34.
Chen, L. & Lee, H. L. (2009), 'Information sharing and order variability control under a
generalized demand model', Management Science 55(5), 781-797.
Chen, M. & Lin, C. (1989), 'Effects of centralization on expected costs in a multi-location
newsboy problem', Journal of the Operational Research Society 40(6), 597-602.
Chen, X. & Zhang, J. (2007), 'Duality approaches to economic lot-sizing games', technical
report, Operations Management Working Papers Series, New York University, May
2007, http://ssrn.com/abstract=1293131, accessed 06/10/2009.
Chen, X. & Zhang, J. (2009), 'A stochastic programming duality approach to inventory
centralization games', Operations Research 57(4), 840-851.
Chen, X. (2009), 'Inventory centralization games with price-dependent demand and
quantity discount', Operations Research 57(6), 1394-1406.
Cheng, F.; Ettl, M.; Lin, G.; Schwarz, M. & Yao, D. (2002), 'Flexible supply contracts via
options', IBM Research Report, Yorktown Heights, NY, December 2002,
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.5.5803&rep=rep1&type=pdf,
accessed 09/10/2010.
Cherikh, M. (2000), 'On the effect of centralization on expected profits in a multi-location
newsboy problem', Journal of the Operational Research Society 51(6), 755-761.
211
Bibliography
Chew, E. K. P.; Lee, L. H. & Lau, Y. L. (2006), 'Component commonality in assembled-

to-stock systems', IIE Transactions 38(3), 239-251.
Chiang, C. & Benton, W. C. (1994), 'Sole sourcing versus dual sourcing under stochastic
demands and lead times', Naval Research Logistics 41(5), 609-624.
Chiang, C. & Chiang, W.-C. (1996), 'Reducing inventory costs by order splitting in the
sole sourcing environment', Journal of the Operational Research Society 47(3), 446-
456.
Chiang, C. (2001), 'Order splitting under periodic review inventory systems', International
Journal of Production Economics 70(1), 67-76.
Chikán, A. (2009), 'An empirical analysis of managerial approaches to the role of
inventories', International Journal of Production Economics 118(1), 131-135.
Ching, W.-K.; Yuen, W.-O. & Loh, A. W. (2003), 'An inventory model with returns and
lateral transshipments', Journal of the Operational Research Society 54(6), 636-641.
Chiou, C.-C. (2008), Transshipment problems in supply chain systems: Review and
extensions, in Kordic, V., ed., Supply Chain: Theory and Applications, I-Tech
Education and Publishing, pp. 427-448.
Chiou, J.-S.; Wu, L.-Y. & Hsu, J. C. (2002), 'The adoption of form postponement strategy
in a global logistics system: The case of Taiwanese information technology industry',
Journal of Business Logistics 23(1), 107-124.
Chmielewicz, K. (1994), Forschungskonzeptionen der Wirtschaftswissenschaft, Schäffer-
Poeschel.
Chod, J. & Rudi, N. (2005), 'Resource flexibility with responsive pricing', Operations
Research 53(3), 532-548.
Cholette, S. (2009), 'Mitigating demand uncertainty across a winery's sales channels
through postponement', International Journal of Production Research 47(13), 3587-
3609.
Chopra, S. & Meindl, P. (2007), Supply Chain Management: Strategy, Planning, and
Operations, Pearson.
Chopra, S. & Sodhi, M. S. (2004), 'Managing risk to avoid supply-chain breakdown', MIT
Sloan Management Review 46(1), 53-62.
Christopher, M. & Peck, H. (2003), Marketing Logistics, Butterworth-Heinemann.
Christopher, M. & Towill, D. (2001), 'An integrated model for the design of agile supply
chains', International Journal of Physical Distribution and Logistics Management
31(4), 235-246.
Christopher, M. (1992), Logistics and Supply Chain Management: Strategies for Reducing
Cost and Improving Service, Pitman.
Christopher, M. (1998), Logistics and Supply Chain Management: Strategies for Reducing
Cost and Improving Service, Financial Times Professional.
Christopher, M. (2000), 'The agile supply chain: Competing in volatile markets', Industrial
Marketing Management 29(1), 37-44.
Cioletti, J. (2007), 'Get the house in order', Beverage World 126(6), 86-89.
Clapham, C. & Nicholson, J. (2009), The Concise Oxford Dictionary of Mathematics,
Oxford University Press.
212
Bibliography
clearpepper supply chain consulting & education (2009), 'Inventory down 80%, sales up
80%. How Sony achieved it', 05/05/2005, http://www.clearpepper.com/supplychain/
sales_up.PDF, accessed 02/18/2009.
CLM (1995), World-Class Logistics: The Challenge of Managing Continuous Change,
Council of Logistics Management.
Cohen, J. (1994), 'The Earth is round (p<.05)', American Psychologist 49(12), 997-1003.
Cohen, M. A.; Hershey, J. C. & Weiss, E. N. (1980), 'Analysis of capacity decisions for
progressive patient care hospital facilities', Health Services Research 15(2), 145-160.
Collier, D. A. (1981), 'The measurement and operating benefits of component part
commonality', Decision Sciences 12(1), 85-96.
Collier, D. A. (1982), 'Aggregate safety stock levels and component part commonality',
Çömez, N.; Stecke, K. E. & Çakanyildirim, M. (2006), 'Virtual pooling considering
transshipment lead time', technical report, School of Management, University of Texas
at Dallas, Richardson, TX, 02/28/2006, http://som.utdallas.edu/faculty/working_papers/
SOM200672.pdf, accessed 03/28/2008.
Cooper, J. C. (1993), 'Logistics strategies for global business', International Journal of
Physical Distribution & Logistics Management 23(4), 12-23.
Cooper, M. C. (1983), 'Freight consolidation and warehouse location strategies in physical
distribution systems', Journal of Business Logistics 4(2), 53-74.
Corbett, C. J. & Rajaram, K. (2004), 'Aggregation of uncertainty and multivariate
dependence: The value of pooling of inventories under non-normal dependent demand',
technical report, UCLA Anderson School of Management, Los Angeles, CA.
Corbett, C. J. & Rajaram, K. (2006), 'A generalization of the inventory pooling effect to
nonnormal dependent demand', Manufacturing & Service Operations Management 8(4),
351-358.
Covino, R. M. (2008), 'Kum & Go's prized prototype', Convenience Store News 44(12),
90-91.
Cox, W. M. & Alm, R. (1998), 'The right stuff: America’s move to mass customization',
Federal Reserve Bank of Dallas 1998 Annual Report, 3-26.
Coyle, J. J.; Bardi, E. J. & Langley, C. J. J. (2003), The Management of Business Logistics:
A Supply Chain Perspective, South-Western Thomson Learning.
Croston, J. D. (1972), 'Forecasting and stock control for intermittent demands',
Operational Research Quarterly 23(3), 289-303.
Croxton, K. L. & Zinn, W. (2005), 'Inventory considerations in network design', Journal of
CSCMP (2010), Council of Supply Chain Management Professionals, 'CSCMP supply
chain management definitions', http://cscmp.org/aboutcscmp/definitions.asp, accessed
01/15/2010.
Daft, R. L. (2009), Organization Theory and Design, South-Western Cengage Learning.
Dai, Y.; Fang, S.-C.; Ling, X. & Nuttle, H. L. W. (2008), 'Risk pooling strategy in a multi-
echelon supply chain with price-sensitive demand', Mathematical Methods of
Operations Research 67(3), 391-421.
213
Bibliography
Dana, J. & Spier, K. (2001), 'Revenue sharing and vertical control in the video rental
industry', The Journal of Industrial Economics 49(3), 223-245.
Dapiran, P. (1992), 'Benetton - global logistics in action', International Journal of Physical
Das, C. & Tyagi, R. (1997), 'Role of inventory and transportation costs in determining the
optimal degree of centralization', Transportation Research: Part E: Logistics and
Das, C. & Tyagi, R. (1999), 'Effect of correlated demands on safety stock centralization:
Patterns of correlation versus degree of centralization', Journal of Business Logistics
20(1), 205-213.
Das, C. (1975), 'Supply and redistribution rules for two-location inventory systems: One-
period analysis', Management Science 21(7), 765-776.
Das, C. (1978), 'A reappraisal of the square root law', International Journal of Physical
Distribution 8(6), 331-336.
Daskin, M. S. (2003), 'Research reports', Industrial Engineer: IE 35(4), 49-50.
Daskin, M. S.; Snyder, L. V. & Berger, R. T. (2005), Chapter 2: Facility location in supply
chain design, in Langevin, A. & Riopel, D., eds., Logistics Systems: Design and
Optimization, Springer, pp. 39-64.
Davila, T. & Wouters, M. (2007), 'An empirical test of inventory, service and cost benefits
from a postponement strategy', International Journal of Production Research 45(10),
2245-2267.
Dekker, R.; Fleischmann, M.; Inderfurth, K. & Van Wassenhove, L. N. (2004), 2
Quantitative models for reverse logistics decision making, in Dekker, R.; Fleischmann,
M.; Inderfurth, K. & Van Wassenhove, L. N., eds., Reverse Logistics: Quantitative
Models for Closed-Loop Supply Chains, Springer, pp. 29-44.
Delfmann, W. (1999), Industrielle Distributionslogistik, in Weber, J. & Baumgarten, H.,
eds., Handbuch Logistik: Management von Material- und Warenflußprozessen,
Schäffer-Poeschel, pp. 181-201.
Delfmann, W. (2000), Logistikkonzeption, Kernelemente in Klaus, P. & Krieger, W., eds.,
Gabler Lexikon Logistik: Management logistischer Netzwerke und Flüsse, Gabler, pp.
322-326.
Denzin, N. K. & Lincoln, Y. S. (2005), The art and practices of interpretation, evaluation,
and presentation, in Denzin, N. K. & Lincoln, Y. S., eds., The SAGE Handbook of
Qualitative Research, Sage, pp. 909-914.
Denzin, N. K. (1978), The Research Act, McGraw-Hill.
Desai, P.; Kekre, S.; Radhakrishnan, S. & Srinivasan, K. (2001), 'Product differentiation
and commonality in design: Balancing revenue and cost drivers', Management Science
47(1), 37-51.
Deshpande, V.; Cohen, M. A. & Donohue, K. (2003), 'A threshold inventory rationing
policy for service-differentiated demand classes', Management Science 49(6), 683-703.
Deuermeyer, B. & Schwarz, L. (1981), A model for the analysis of system service-level in
warehouse-retailer distribution systems: The identical retailer case in Schwarz, L. B.,
ed., Multi-Level Production/Inventory Control Systems: Theory and Practice, North-
Holland, pp. 163-193.
214
Bibliography
Diks, E. B. & de Kok, A. G. (1996), 'Controlling a divergent 2-echelon network with

transshipments using the consistent appropriate share rationing policy', International
Journal of Production Economics 45(1-3), 369-379.
Dilts, D. M. (2005), 'Risk mitigation in supply chain management: Cost of SC disruptions
& strategies to reduce or hedge uncertainty', Professor of Operations Management,
Owen Graduate School of Management, Vanderbilt University, Nashville, TN, February
2005, http://www2.owen.vanderbilt.edu/david.dilts/Mgt472_SCM/Mgt472-11-Risk_Mi
tigation.ppt, accessed 07/22/2008.
Diruf, G. (2005), 'Risk-Pooling-Strategien in der Modedistribution', technical report,
Bamberger betriebswirtschaftliche Beiträge 138, Otto-Friedrich Universität Bamberg,
Lehrstuhl für Produktion und Logistik.
Diruf, G. (2007), Risk-Pooling-Kooperationen im Modehandel: Erfolgsfaktoren und
Verbesserungspotenziale, in Vahrenkamp, R. & Siepermann, C., eds., Risikomanagment
in der Supply Chain: Gefahren abwehren, Chancen nutzen, Erfolg generieren, Erich
Schmidt Verlag, pp. 235-252.
Dobos, I. & Richter, K. (2000), 'The integer EOQ repair and waste disposal model–further
analysis', Central European Journal of Operations Research 8(2), 173-194.
Doch, S. A. (2009), Logistische Leistungsdifferenzierung im Supply Chain Management,
Universitätsverlag der TU Berlin.
Dodge, Y., ed. (2006), The Oxford Dictionary of Statistical Terms, Oxford University
Press.
Dogramaci, A. (1979), 'Design of common components considering implications of
inventory costs and forecasting', AIIE Transactions 11(2), 129-135.
Dominguez, H. & Lashkari, R. S. (2004), 'Model for integrating the supply chain of an
appliance company: A value of information approach', International Journal of
Production Research 42(11), 2113-2140.
Domschke, W. & Drexl, A. (1996), Logistik: Standorte, Oldenbourg.
Donaldson, L. (1996), For Positivist Organization Theory, Sage.
Donaldson, L. (1999), Chapter 2: The normal science of structural contingency theory, in
Clegg, S. R. & Hardy, C., eds., Studying Organization: Theory & Method, Sage, pp. 51-
70.
Donaldson, L. (2001), The Contingency Theory of Organizations, Sage.
Dong, L. & Rudi, N. (2002), 'Supply chain interaction under transshipments: Exogenous
vs. endogenous wholesale prices', technical report, Olin School of Business,
Washington University, St. Louis, MO.
Dong, L. & Rudi, N. (2004), 'Who benefits from transshipment? Exogenous vs.
endogenous wholesale prices', Management Science 50(5), 645-657.
Donnellan, B.; Larsen, T.; Levine, L. & DeGross, J., eds. (2006), The Transfer and
Diffusion of Information Technology for Organizational Resilience: IFIP TC8 WG 8.6
International Working Conference, June 7-10, 2006, Galway, Ireland, Springer.
Dorofeev, S. & Grant, P. (2006), Statistics for Real-Life Sample Surveys: Non-Simple-
Random Samples and Weighted Data, Cambridge University Press.
Drazin, R. & van de Ven, A. H. (1985), 'Alternative form of strategic fit in contingency
theory', Administrative Science Quarterly 30(4), 514-539.
215
Bibliography
Drechsel, J. & Kimms, A. (2007), Berechnung von Kernelementen in kooperativen

Bestellmengen-Spielen, in Koschke, R.; Herzog, O.; Rödiger, K.–H. & Ronthaler, M.,
eds., 'Lecture Notes in Informatics - Informatik 2007 Proceedings "Informatik trifft
Logistik"', pp. 79-82.
Dröge, C. L.; Germain, R. N. & Spears, N. (1995), Form postponement as a strategic
initiative affecting organizational design, in 'Proceedings of the American Marketing
Association Summer Educators' Conference', pp. 163-169.
Dror, M. & Hartman, B. C. (2007), 'Shipment consolidation: Who pays for it and how
much?', Management Science 53(1), 78-87.
Dror, M.; Guardiola, L.; Meca, A. & Puerto, J. (2008), 'Dynamic realization games in
newsvendor inventory centralization', International Journal of Game Theory 37(1), 139-
153.
Dubelaar, C.; Chow, G. & Larson, P. D. (2001), 'Relationships between inventory, sales
and service in a retail chain store operation', International Journal of Physical
Duller, C. (2008), Einführung in die nichtparametrische Statistik mit SAS und R: Ein
anwendungsorientiertes Lehr- und Arbeitsbuch, Physica Verlag.
Durand, B. & Paché, G. (2006), From traditional retailing to e-tailing: The death and
rebirth of the hypermarket format?, in Seppä, M. et al., eds., 'Conference Proceedings of
the 5th Frontiers of e-Business Research', Tampere University of Technology (TUT)
and University of Tampere (UTA), pp. 95-108.
Durand, B. (2007), Warehousing strategy: The limits of the theory of inventory
centralization, in '42nd Canadian Transportation Research Forum', Winnipeg, June 3-6,
2007.
Duvall, M. (2000), 'Cost-saving idea: Virtual warehouses', Inter@ctive Week 7(42), 18.
Dzimbiri, L. B. (2009), Organisation and Management Theories: An African Focus,
Cuvillier Verlag.
Electrical Wholesaling (1997), 'Trade Service creates virtual warehouse', Electrical
Wholesaling 78(11), 11.
Electrical Wholesaling (2003), 'Distributors take advantage of immediate stock e-virtual
warehouse', Electrical Wholesaling 84(9), 19.
ElMaraghy, H. A. & Mahmoudi, N. (2009), 'Concurrent design of product modules
structure and global supply chain configurations', International Journal of Computer
Integrated Manufacturing 22(6), 483-493.
Enarsson, L. (2006), Future Logistics Challenges, Copenhagen Business School Press.
Eppen, G. & Schrage, L. (1981), Centralized ordering policies in a multi-warehouse
system with lead times and random demand, in Schwarz, L. B., ed., Multi-Level
Production/Inventory Systems: Theory and Practice, North-Holland, pp. 51-67.
Eppen, G. D. (1979), 'Effects of centralization on expected costs in a multi-location
newsboy problem', Management Science 25(5), 498-501.
Erkip, N.; Hausman, W. H. & Nahmias, S. (1990), 'Optimal centralized ordering policies in
multi-echelon inventory systems with correlated demands', Management Science 36(3),
381-392.
216
Bibliography
Ernst, R. & Kamrad, B. (2000), 'Evaluation of supply chain structures through

modularization and postponement', European Journal of Operational Research 124(3),
495-510.
Ernst, R. & Kouvelis, P. (1999), 'The effects of selling packaged goods on inventory
decisions', Management Science 45(8), 1142-1155.
Evers, P. T. & Beier, F. J. (1993), 'The portfolio effect and multiple consolidation points:
A critical assessment of the square root law', Journal of Business Logistics 14(2), 109-
125.
Evers, P. T. & Beier, F. J. (1998), 'Operational aspects of inventory consolidation decision
making', Journal of Business Logistics 19(1), 173-189.
Evers, P. T. (1994), 'The occurrence of statistical economies of scale in intermodal
transportation', Transportation Journal 33(4), 51-63.
Evers, P. T. (1995), 'Expanding the square root law: An analysis of both safety and cycle
stocks', Logistics and Transportation Review 31(1), 1-20.
Evers, P. T. (1996), 'The impact of transshipments on safety stock requirements', Journal
of Business Logistics 17(1), 109-133.
Evers, P. T. (1997), 'Hidden benefits of emergency transshipments', Journal of Business
Logistics 18(2), 55-76.
Evers, P. T. (1999), 'Filling customer orders from multiple locations: A comparison of
pooling methods', Journal of Business Logistics 20(1), 121-139.
Evers, P. T. (2001), 'Heuristics for assessing emergency transshipments', European Journal
of Operational Research 129(2), 311-316.
Eynan, A. & Fouque, T. (2003), 'Capturing the risk-pooling effect through demand
reshape', Management Science 49(6), 704-717.
Eynan, A. & Fouque, T. (2005), 'Benefiting from the risk-pooling effect: Internal
(component commonality) vs. external (demand reshape) efforts', International Journal
of Services and Operations Management 1(1), 90-99.
Eynan, A. & Rosenblatt, M. J. (1996), 'Component commonality effects on inventory
costs', IIE Transactions 28(2), 93-104.
Eynan, A. (1996), 'The impact of demands' correlation on the effectiveness of component
commonality', International Journal of Production Research 34(6), 1581-1602.
Eynan, A. (1999), 'The multi-location inventory centralization problem with first-come,
first-serve allocation', European Journal of Operational Research 114(1), 38-49.
Faloon, K. (1999), 'ARWI launches virtual warehouse', Supply House Times 42(9), 30.
Fawcett, P.; McLeish, R. E. & Ogden, I. D. (1992), Logistics Management, Pitman.
Feitzinger, E. & Lee, H. L. (1997), 'Mass customization at Hewlett-Packard: The power of
postponement', Harvard Business Review 75(1), 116-121.
Fielding, N. & Fielding, J. L. (1986), Linking Data: The Articulation of Qualitative and
Quantitative Methods in Social Research, Sage.
Fisher, M. & Raman, A. (1996), 'Reducing the cost of demand uncertainty through
accurate response to early sales', Operations Research 44(1), 87-99.
Flaks, M. (1967), Total cost approach to physical distribution, in Marks, N. E. & Taylor, R.
M., eds., Marketing Logistics: Perspectives and Viewpoints, Wiley, pp. 264-269.
217
Bibliography
Fleisch, E.; Österle, H. & Powell, S. (2004), 'Rapid implementation of enterprise resource
planning systems', Journal of Organizational Computing & Electronic Commerce
14(2), 107-126.
Fleischmann, M.; Bloemhof-Ruwaard, J. M.; Beullens, P. & Dekker, R. (2004), 4 Reverse
logistics network design, in Dekker, R.; Fleischmann, M.; Inderfurth, K. & Van
Wassenhove, L. N., eds., Reverse Logistics: Quantitative Models for Closed-Loop
Supply Chains, Springer, pp. 65-94.
Fliedner, G. & Vokurka, R. J. (1997), 'Agility: Competitive weapon of the 1990s and
beyond?', Production and Inventory Management Journal 38(3), 19-24.
Flynn, B. B.; Sakakibara, S.; Schroeder, R. G.; Bates, K. A. & Flynn, E. J. (1990),
'Empirical research methods in operations management', Journal of Operations
Fong, D. K. H. & Gempeshaw, V. M. (1996), 'Evaluating re-order levels and probabilities
of stockout during a lead time for some stock control models', Journal of Operational
Research Society 47(2), 321-328.
Fong, D. K. H. & Ord, K. J. (1993), 'Estimating moments of the effective lead time for a
stock control mode with independent normal lead time', Journal of Operational
Fong, D. K. H. (1992), 'A note on exact moment computation for normal lead times in the
two-supplier case', Journal of Operational Research Society 43(1), 63-69.
Fong, D. K. H.; Gempeshaw, V. M. & Ord, K. J. (2000), 'Analysis of a dual sourcing
inventory model with normal unit demand and Erlang mixture lead times', European
Journal of Operational Research 120(1), 97-107.
Fowler, F. (1993), Survey Research Methods, Sage.
FPT (2007a), 'For Paper Technology: Die Sicht der Papierindustrie: FPT Druckstudie –
Teil II', technical report, Forschungsvereinigung Papiertechnik e.V. (FPT) c/o
Papiertechnische Stiftung (PTS), Munich, 10/16/2007.
FPT (2007b), 'For Paper Technology: FPT-Workshop Nr. 4: Zukunft der Grafischen
Papiere - Wie können Papiererzeuger und deren Zulieferer mit dem Papiergroßhandel
und den Druckern über eine optimierte Sortenpolitik zu einer verbesserten
Wertschöpfung gelangen?', technical report, Forschungsvereinigung Papiertechnik e.V.
(FPT) c/o Papiertechnische Stiftung (PTS), Munich, 10/24/2007.
Franse, K. (1999), 'D&H unveils e-commerce division', VARBusiness 15(7), 20.
Frese, E. (1992), Organisationstheorie: Historische Entwicklung - Ansätze - Perspektiven,
Gabler.
Frese, E. (2000), Grundlagen der Organisation, Gabler.
Friedman, J. P., ed. (2007), Dictionary of Business Terms, Barron's Educational Series.
Fritsch, M.; Wein, T. & Ewers, H.-J. (2001), Marktversagen und Wirtschaftspolitik:
Mikroökonomische Grundlagen staatlichen Handelns, Vahlen.
Frontline Solutions Europe (2002), 'Walk through the virtual warehouse', Frontline
Solutions Europe 11(6), 38.
Fung, S. H.; Cheung, C. F.; Lee, W. B. & Kwok, S. K. (2005), 'A virtual warehouse system
for production logistics', Production Planning & Control 16(6), 597-607.
218
Bibliography
Ganesh, M.; Raghunathan, S. & Rajendran, C. (2008), 'The value of information sharing in
a multi-product supply chain with product substitution', IIE Transactions 40(12), 1124-
1140.
Ganeshan, R.; Ring, L. J. & Strong, J. S. (2007), 'A mathematical approach to designing
central vs. local ordering in retail', International Journal of Production Economics
108(1-2), 341-348.
Ganeshan, R.; Tyworth, J. E. & Guo, Y. (1999), 'Dual sourced supply chains: The discount
supplier option', Transportation Research: Part E: Logistics and Transportation Review
35(1), 11-23.
García-Dastugue, S. J. & Lambert, D. M. (2007), 'Interorganizational time-based
postponement in the supply chain', Journal of Business Logistics 28(1), 57-81.
Garg, A. & Tang, C. S. (1997), 'On postponement strategies for product families with
multiple points of differentiation', IIE Transactions 29(8), 641-650.
Garson, D. (2009), 'Chi-square significance tests', North Carolina State University,
Raleigh, NC, http://faculty.chass.ncsu.edu/garson/PA765/chisq.htm, accessed 10/24/
2009.
Gaukler, G. M. (2007), 'Supply chain concepts: Double marginalization and risk pooling',
Department of Industrial and Systems Engineering, Texas A&M University, College
Station, TX, 11/14/2007, http://ise.tamu.edu/people/faculty/gaukler/scm.pdf, accessed
08/05/2008.
Gaur, S. & Ravindran, A. R. (2006), 'A bi-criteria model for the inventory aggregation
problem under risk pooling', Computers & Industrial Engineering 51(3), 482-501.
Geetha, K. K. & Achary, K. K. (2000), 'Are more suppliers better? Generalizing the Guo
and Ganeshan procedure', Journal of the Operational Research Society 51(10), 1179-
1181.
Gerchak, Y. & Gupta, D. (1991), 'On apportioning costs to customers in centralized
continuous review inventory systems', Journal of Operations Management 10(4), 546-
551.
Gerchak, Y. & He, Q.-M. (2003), 'On the relation between the benefits of risk pooling and
the variability of demand', IIE Transactions 35(11), 1027-1032.
Gerchak, Y. & Henig, M. (1986), 'An inventory model with component commonality',
Operations Research Letters 5(3), 157-160.
Gerchak, Y. & Henig, M. (1989), 'Component commonality in assemble-to-order systems:
Models and properties', Naval Research Logistics 36(1), 61-68.
Gerchak, Y. & Mossman, D. (1992), 'On the effect of demand randomness on inventories
and costs', Operations Research 40(4), 804-808.
Gerchak, Y.; Magazine, M. J. & Gamble, A. B. (1988), 'Component commonality with
service level requirements', Management Science 34(6), 753-760.
Gerchak, Y.; Tripathy, A. & Wang, K. (1996), 'Co-production models with random
functionality yields', IIE Transactions 28(5), 391-403.
Ghemawat, P. & Nueno, J. L. (2003), Zara: Fast Fashion, Harvard Business School
Publishing.
Ghiani, G.; Laporte, G. & Musmanno, R. (2004), Introduction to Logistics Systems and
Planning Control, Wiley.
219
Bibliography
Ghobbar, A. A. & Friend, C. H. (2003), 'Evaluation of forecasting methods for intermittent

parts demand in the field of aviation: A predictive model', Computers and Operations
Research 30(14), 2097-2114.
Ghodsypour, S. H. & O'Brien, C. (2001), 'The total cost of logistics in supplier selection,
under conditions of multiple sourcing, multiple criteria and capacity constraint',
International Journal of Production Economics 73(1), 15-27.
Gibson, G. & Horner, T. (2005), 'JDEtips on manufacturing: JD Edwards forecasting',
JDEtips Journal VI(6), 1-10.
Gilmore, J. H. & Pine II, B. J. (1997), 'The four faces of mass customization', Harvard
Business Review 75(1), 91-101.
Gobsch, B. (2007), Deterministische Losgrößenmodelle der Entsorgungslogistik und
Reverse Logistics, Wissenschaftlicher Verlag Berlin.
Goldsby, T. J. & Stank, T. P. (2000), 'World class logistics performance and
environmentally responsible logistics practices', Journal of Business Logistics 21(2),
187-208.
Gong, Y. & Yucesan, E. (2006), 'The multi-location transshipment problem with positive
replenishment lead times', Research Paper ERS-2006-048-LIS Revision, Erasmus Re-
search Institute of Management (ERIM), RSM Erasmus University, Rotterdam,
09/07/2006, http://publishing.eur.nl/ir/repub/asset/7947/ERS-2006-048-LIS.pdf, access
08/02/2008.
Gourmet Retailer (2002), 'Web site news', Gourmet Retailer 23(2), 12.
Goyal, M. & Netessine, S. (2005), 'Capacity investment and the interplay between volume
flexibility and product flexibility', technical report, The Wharton School, University of
Pennsylvania, Philadelphia, PA, May 2005, http://opimweb.wharton.upenn.edu/docu
ments/research/Goyal%20Netessine%20b.pdf, accessed 09/10/2010.
Goyal, M. & Netessine, S. (2006), 'Strategic technology choice and capacity investment
under demand uncertainty', working paper, The Wharton School, University of
Pennsylvania, Philadelphia, PA, April 2006, http://www.rhsmith.umd.edu/faculty/mgo
yal/Goyal%20Netessine_Tech-Choice.pdf, accessed 09/20/2009.
Goyal, M.; Netessine, S. & Randall, T. (2006), 'Deployment of manufacturing flexibility:
An empirical analysis of the North American automotive industry', technical report,
Robert H. Smith School of Business, University of Maryland, College Park, MD, March
2006, http://opimweb.wharton.upenn.edu/documents/research/GoyalNetessineRandall.
pdf, accessed 09/11/2009.
Gradshteyn, I. S. & Ryzhik, M. (2007), Table of Integrals, Series and Products, Academic
Press.
Grahovac, J. & Chakravarty, A. (2001), 'Sharing and lateral transshipment of inventory in
a supply chain with expensive low-demand items', Management Science 47(4), 580-594.
Graman, G. A. & Magazine, M. J. (1998), An analysis of packaging postponement, in
'MSOM Conference Proceedings', University of Washington, Seattle, WA.
Graman, G. A. & Magazine, M. J. (2006), 'Implementation issues influencing the decision
to adopt postponement', International Journal of Operations & Production
Management 26(10), 1068-1083.
Graman, G. A. & Sanders, N. R. (2009), 'Modelling the tradeoff between postponement
capacity and forecast accuracy', Production Planning & Control 20(3), 206-215.
220
Bibliography
Graman, G. A. (2010), 'A partial-postponement decision cost model', European Journal of

Granot, D. & Sošić, G. (2003), 'A three-stage model for a decentralized distribution system
of retailers', Operations Research 51(5), 771-784.
Granot, D. & Yin, S. (2008), 'Price and order postponement in a decentralized newsvendor
model with multiplicative and price-dependent demand', Operations Research 56(1),
121-139.
Graves, S. C. (2008), Chapter 3: Flexibility principles, in Chhajed, D. & Lowe, T. J., eds.,
Building Intuition: Insights from Basic Operations Management Models and Principles,
Gravetter, F. J. & Wallnau, L. B. (2008), Statistics for the Behavioral Sciences, Cengage
Learning.
Griffis, S. E.; Goldsby, T. J. & Cooper, M. (2003), 'Web-based and mail surveys: A
comparison of response, data, and cost', Journal of Business Logistics 24(2), 237-258.
Grotzinger, S. J.; Srinivasan, R.; Akella, R. & Bollapragada, S. (1993), 'Component
procurement and allocation for products assembled to forecast: Risk and pooling
effects', IBM Journal of Research and Development 37(4), 523-536.
Groves, R. M.; Fowler Jr., F. J.; Couper, M. P.; Lepkowski, J. M.; Singer, E. &
Tourangeau, R. (2004), Survey Methodology, John Wiley & Sons.
Grün, O.; Jammernegg, W. & Kummer, S. (2009), Grundzüge der Beschaffung,
Produktion und Logistik, Pearson Studium.
Gürbüz, M. Ç.; Moinzadeh, K. & Zhou, Y.-P. (2007), 'Coordinated replenishment
strategies in inventory/distribution systems', Management Science 53(2), 293-307.
Guerrero, H. H. (1985), 'The effect of various production strategies on product structures
with commonality', Journal of Operations Management 5(4), 395-410.
Güttler, K. (2009), Formale Organisationsstrukturen in wachstumsorientierten kleinen und
mittleren Unternehmen, Gabler.
Guglielmo, C. (1999), 'IQVC builds virtual warehouse', Inter@ctive Week 6(25), 37.
Guo, Y. & Ganeshan, R. (1995), 'Are more suppliers better?', Journal of the Operational
Gupta, D. & Benjaafar, S. (2004), 'Make-to-order, make-to-stock, or delay product
differentiation? A common framework for modeling and analysis', IIE Transactions
36(6), 529-546.
Gupta, O. K. & Kini, R. B. (1995), 'Is price-quantity discount dead in a just-in-time
environment?', International Journal of Operations and Production Management 15(9),
261-270.
Gutenberg, E. (1983), Grundlagen der Betriebswirtschaftslehre, Band 1, Die Produktion,
Springer.
Gutierrez, R. S.; Solis, A. O. & Mukhopadhyay, S. (2008), 'Lumpy demand forecasting
using neural networks', International Journal of Production Economics 111(2), 409-
420.
Haimes, Y. Y. (2009), Risk Modeling, Assessment, and Management, John Wiley & Sons.
221
Bibliography
Hall, R. W. (2004), 'Domicile selection and risk pooling for trucking networks', IIE
Transactions 36(4), 299-305.
Hammel, T.; Phelps, T. & Kuettner, D. (2002), 'The re-engineering of Hewlett-Packard's
CD-RW supply chain', Supply Chain Management: An International Journal 7(3), 113-
118.
Hamstra, M. (2009), 'Retailers increase focus on SKU rationalization', SN: Supermarket
News 57(41), 28-30.
Harper, R. (2008), 'Silver medal: Ready Pac', SN: Supermarket News 56(40), 54.
Harris, R. G. (1977), 'Economies of traffic density in the rail freight industry', Bell Journal
of Economics 8(2), 556-564.
Harrison, A. & Skipworth, H. (2008), 'Implications of form postponement to
manufacturing: A cross case comparison', International Journal of Production Research
46(1), 173-195.
Hartman, B. & Dror, M. (1996), 'Cost allocation in continuous-review inventory models',
Naval Research Logistics 43(4), 549-561.
Hartman, B. C. & Dror, M. (2003), 'Optimizing centralized inventory operations in a
cooperative game theory setting', IIE Transactions 35(3), 243-258.
Hartman, B. C. & Dror, M. (2005), 'Allocation of gains from inventory centralization in
newsvendor environments', IIE Transactions 37(2), 93-107.
Hartman, B. C.; Dror, M. & Shaked, M. (2000), 'Cores of inventory centralization games',
Games and Economic Behavior 31(1), 26-49.
Hayes, R. & Wheelwright, S. (1984), Restoring Our Competitive Edge: Competing
through Manufacturing, John Wiley & Sons.
Hayya, J. C.; Christy, D. P. & Pan, A. (1987), 'Reducing inventory uncertainty: A reorder
point system with two vendors', Production and Inventory Management 28(2), 43-48.
Hearn, G. (2007), 'Shred or dread', Airfinance Journal April 1, 2007 (299), 9.
Heese, H. S. & Swaminathan, J. M. (2006), 'Product line design with component
commonality and cost-reduction effort', Manufacturing & Service Operations
Heil, U. (2006), Kooperative Risk-Pooling-Strategien im Modehandel, Lang Verlag.
Hempel, C. G. (1970), Methods of concept formation in science, in Neurath, O.; Carnap,
R. & Morris, C., eds., Formations of the Unity of Science, University of Chicago Press.
Hendricks, K. B. & Singhal, V. R. (2002), 'How supply chain glitches torpedo shareholder
value', Supply Chain Management Review 6(1), 18-24.
Hendricks, K. B. & Singhal, V. R. (2003), 'The effect of supply chain glitches on
shareholder wealth', Journal of Operations Management 21(5), 501-522.
Hendricks, K. B. & Singhal, V. R. (2005a), 'Association between supply chain glitches and
operating performance', Management Science 51(5), 695-711.
Hendricks, K. B. & Singhal, V. R. (2005b), 'Supply-chain disruptions: Torpedo
shareholder value and profitability', Metal Producing & Processing 43(6), 36-35.
Hendricks, K. B. & Singhal, V. R. (2005c), 'An empirical analysis of the effect of supply
chain disruptions on long-run stock price performance and equity risk of the firm',
Production & Operations Management 14(1), 35-52.
222
Bibliography
Henig, M.; Gerchak, Y.; Ernst, R. & Pyke, D. (1997), 'An inventory model embedded in
designing a supply contract', Management Science 43(2), 184-189.
Herer, Y. T. & Rashit, A. (1999), 'Lateral stock transshipments in a two-location inventory
system with fixed and joint replenishment costs', Naval Research Logistics 46(5), 525-
547.
Herer, Y. T. & Tzur, M. (2003), 'Optimal and heuristic algorithms for the multi-location
dynamic transshipment problem with fixed transshipment costs', IIE Transactions 35(5),
419-432.
Herer, Y. T.; Tzur, M. & Yücesan, E. (2002), 'Transshipments: An emerging inventory
recourse to achieve supply chain leagility', International Journal of Production
Economics 80(3), 201-213.
Herer, Y. T.; Tzur, M. & Yücesan, E. (2006), 'The multilocation transshipment problem',
IIE Transactions 38(3), 185-200.
Herron, D. P. (1987), 'Integrated inventory management', Joumal of Business Logistics
8(1), 96-116.
Heskett, J. L. & Signorelli, S. (1984), Benetton (A), Harvard teaching case 9-685-014,
Harvard Business Publishing.
Heskett, J. L.; Glaskowsky, N. A. J. & Ivie, R. M. (1974), Business Logistics, Ronald
Press.
Hiddemann, T. (2007), Operatives Management und der Erfolg junger, innovativer
Unternehmen: Die moderierende Wirkung der externen und internen Unsicherheit,
DUV.
Hill, R. M. (1996), 'Order splitting in continuous review (Q, r) inventory models',
European Journal of Operational Research 95(1), 53-61.
Hillier, M. S. (2002a), 'Using commonality as backup safety stock', European Journal of
Hillier, M. S. (2002b), 'The costs and benefits of commonality in assemble-to-order
systems with a (Q,r)-policy for component replenishment', European Journal of
Hilmola, O.-P. & Szekely, B. (2006), 'Logistics development in Finnish and Swedish
companies with respect of Russia and four Asian countries: Traffic flow and
warehousing analysis from current situation and likely development trends', Research
Report 175, Lappeenranta University of Technology, Department of Industrial
Engineering and Management, Kouvola Research Unit, https://oa.doria.fi/bitstream/han
dle/10024/31039/TMP.objres.384.pdf?sequence=1, accessed 09/10/2010.
Ho, T.-H.; Lim, N. & Cui, T. H. (2009), 'Is inventory centralization profitable? An
experimental investigation', technical report, Haas School of Business, University of
California at Berkeley; Bauer College of Business, University of Houston; Carlson
School of Management, University of Minnesota, 01/31/2009, http://faculty.haas.berke
ley.edu/hoteck/PAPERS/Risk.pdf, accessed 06/10/2009.
Hoffman, W. (2005), 'When supply chains shatter', Traffic World 269(26), 13.
Hoffman, W. (2008), 'Taking less stock', Journal of Commerce (15307557) 9(17), 44.
223
Bibliography
Höhne, E. (2009), Kontingenztheorie in Schwaiger, M. & Meyer, A., eds., Theorien und
Methoden der Betriebswirtschaft: Handbuch für Wissenschaftler und Studierende,
Vahlen, pp. 83-96.
Holtmann, D. (2007), Grundlegende multivariate Modelle der sozialwissenschaftlichen
Datenanalyse, Universitätsverlag Potsdam.
Hong, J. D. & Hayya, J. C. (1992), 'Just-in-time purchasing: Single or multiple sourcing?',
Hong, J.-D.; Hayya, J. C. & Kim, S.-L. (1992), 'JIT purchasing and setup reduction in an
integrated inventory model', International Journal of Production Research 30(2), 255-
266.
Hong-Minh, S. M.; Disney, S. M. & Naim, M. M. (2000), 'The dynamics of emergency
transhipment supply chains', International Journal of Physical Distribution & Logistics
Management 30(9), 788-816.
Hoppe, B. & Tardos, E. (2000), 'The quickest transshipment problem', Mathematics of
Operations Research 25(1), 36-62.
Howard, K. A. (1994), 'Postponement of packaging and product differentiation for lower
logistics costs', Journal of Electronics Manufacturing 4(2), 65-69.
Hsu, A. & Bassok, Y. (1999), 'Random yield and random demand in a production system
with downward substitution', Operations Research 47(2), 277-290.
HTT (2005), 'Pier 1 talks of plans for turnaround', Home Textiles Today 26(19), 4.
Hu, J.; Watson, E. & Schneider, H. (1997), Centralized versus individual ordering in
multilocation inventory systems, in '1997 Proceedings - Decision Sciences Institute
Annual Meeting', pp. 1486-1488.
Hu, J.; Watson, E. & Schneider, H. (2005), 'Approximate solutions for multi-location
inventory systems with transshipments', International Journal of Production Economics
97(1), 31-43.
Hu, X.; Duenyas, I. & Kapuscinski, R. (2007), 'Existence of coordinating transshipment
prices in a two-location inventory model', Management Science 53(8), 1289-1302.
Hu, X.; Duenyas, I. & Kapuscinski, R. (2008), 'Optimal joint inventory and transshipment
control under uncertain capacity', Operations Research 56(4), 881-897.
Huang, C. W. & Lo, C. P. (2003), 'Using postponed manufacturing to reconfigure the
supply chain in the desktop personal company industry: The case Taiwan', International
Journal of Management 20(2), 241-256.
Huang, G. Q.; Zhang, X. Y. & Lo, V. H. Y. (2005), 'Optimal supply chain configuration
for platform products: Impacts of commonality, demand variability and quantity
discount', International Journal of Mass Customisation 1(1), 107-133.
Huang, X.; Gattiker, T. F. & Schroeder, R. G. (2008), 'Structure–infrastructure alignment:
The relationship between TQM orientation and the adoption of supplier-facing
electronic commerce among manufacturers', Journal of Supply Chain Management: A
Global Review of Purchasing & Supply 44(1), 40-54.
Huang, Y.-Y. & Li, S.-J. (2008a), 'Suitable application situations of different postpone-
ment approaches: Standardization vs. modularization', Journal of Manufacturing Sys-
tems 27(3), 111-122.
224
Bibliography
Huang, Y.-Y. & Li, S.-J. (2008b), 'Postponement application in Greater China and its
related determinants', Transportation Journal 47(3), 5-21.
Hubbard, D. (2009), The Failure of Risk Management: Why It's Broken and How to Fix It,
John Wiley & Sons.
Hubbard, D. (2010), How to Measure Anything: Finding the Value of "Intangibles" in
Business, John Wiley & Sons.
Hult, G. T. M.; Ketchen, D. J. & Arrfelt, M. (2007), 'Strategic supply chain management:
Improving performance through a culture of competitiveness and knowledge
development', Strategic Management Journal 28(10), 1035-1052.
Humair, S. & Willems, S. P. (2006), 'Optimizing strategic safety stock placement in supply
chains with clusters of commonality', Operations Research 54(4), 725-742.
Hutchings, J. (1999), 'Internet taps into the horizontal supply chain', Modern Power
Systems 19(11), 49.
Hwarng, H. B.; Chong, C. S. P.; Xie, N. & Burgess, T. F. (2005), 'Modelling a complex
supply chain: Understanding the effect of simplified assumptions', International Journal
of Production Research 43(13), 2829-2872.
Ihde, G. B. (1976), Größenersparnisse in der Distribution, Gabler.
Ihde, G. B. (2001), Transport, Verkehr, Logistik: Gesamtwirtschaftliche Aspekte und ein-
zelwirtschaftliche Handhabung, Vahlen.
IMF (2009), 'World Economic Outlook Database, April 2009: Nominal GDP list of
countries. Data for the year 2008', International Monetary Fund, Washington, D.C.
Inderfurth, K. (1991), 'Safety stock optimization in multi-stage inventory systems',
International Journal of Production Economics 24(1-2), 103-113.
Iyer, A. V. & Jain, A. (2004), 'Modeling the impact of merging capacity in production-
inventory systems', Management Science 50(8), 1082-1094.
Jabbonsky, L. (1994), 'Stop SKU-ing around', Beverage World 113(1574), 6.
Jackson, G. C. (1985), 'A survey of freight consolidation practices', Journal of Business
Logistics 6(1), 13-34.
Jackson, P. & Muckstadt, J. (1989), 'Risk pooling in a two-period, two-echelon inventory
stocking and allocation problem', Naval Research Logistics 36, 1-26.
Jackson, P. L. & Muckstadt, J. A. (1984a), 'A two-period, two-echelon inventory stocking
and allocation problem', Technical Report 616, School of Operations Research and
Industrial Engineering, Cornell University, Ithaca, NY.
Jackson, P. L. & Muckstadt, J. A. (1984b), 'Risk pooling in a two-period, two-echelon
inventory stocking and allocation problem', Technical Report 634, School of Operations
Research and Industrial Engineering, Cornell University, Ithaca, NY.
Jackson, P. L. (1988), 'Stock allocation in a two-echelon distribution system or "what to do
until your ship comes in"', Management Science 34(7), 880-895.
Jain, A. (2007), 'Value of capacity pooling in supply chains with heterogeneous customers',
Jain, N. & Paul, A. (2001), 'A generalized model of operations reversal for fashion goods',
225
Bibliography
Jans, R.; Degraeve, Z. & Schepens, L. (2008), 'Analysis of an industrial component

commonality problem', European Journal of Operational Research 186(2), 801-811.
Janssen, F.; de Kok, T. & van der D Schouten, F. (2000), 'Approximate analysis of the
delivery splitting model', Journal of the Operational Research Society 51(10), 1136-
1147.
Jayaraman, V. (1998), 'Transportation, facility location and inventory issues in distribution
network design', International Journal of Operations & Production Management 18(5),
471-494.
Jiang, K.; Lee, H. L. & Seifert, R. W. (2006), 'Satisfying customer preferences via mass
customization and mass production', IIE Transactions 38(1), 25-38.
Jick, T. D. (1979), 'Mixing qualitative and quantitative methods: Triangulation in action',
Administrative Science Quarterly 24(4), 602-611.
Johnson, E. A. & Anderson, E. (2000), 'Postponement strategies for channel derivatives',
Johnson, K. L. (2005), 'Inventory modeling', Intel Technology Journal 9(3), 233-238.
Johnson, P. F.; Leenders, M. R. & Fearon, H. E. (2006), 'Supply's growing status and
influence: A sixteen-year perspective', Journal of Supply Chain Management: A Global
Review of Purchasing & Supply 42(2), 33-43.
Johnston, F. R. & Boylan, J. E. (1996), 'Forecasting for items with intermittent demand',
Journal of the Operational Research Society 47(1), 113-121.
Jönsson, H. & Silver, E. A. (1987a), 'Analysis of a two-echelon inventory control system
with complete redistribution', Management Science 33(2), 215-227.
Jönsson, H. & Silver, E. A. (1987b), 'Stock allocation among a central warehouse and
identical regional warehouses in a particular push inventory control system',
International Journal of Production Research 25(2), 191-205.
Jönsson, H.; Jörnsten, K. & Silver, E. A. (1993), 'Application of the scenario aggregation
approach to a two-stage, stochastic, common component, inventory problem with a
budget constraint', European Journal of Operational Research 68(2), 196-211.
Jordan, W. C. & Graves, S. C. (1995), 'Principles on the benefits of manufacturing process
flexibility', Management Science 41(4), 577-594.
Jorion, P. (2009), Financial Risk Manager Handbook, Wiley.
Jung, B.-R.; Sun, B.-G.; Kim, J.-S. & Ahn, S.-E. (2003), 'Modeling lateral transshipments
in multiechelon repairable-item inventory systems with finite repair channels',
Computers & Operations Research 30(9), 1401-1418.
Kang, H. (1986), 'Unstable weights in the combination of forecasts', Management Science
32(6), 683-695.
Kapuscinski, R. & Tayur, S. (1999), 'Variance vs. standard deviation: Variability reduction
through operations reversal', Management Science 45(5), 765-767.
Kelle, P. & Miller, P. A. (2001), 'Stockout risk and order splitting', International Journal of
Production Economics 71(1), 407-415.
Kelle, P. & Silver, E. A. (1990a), 'Safety stock reduction by order splitting', Naval
Research Logistics 37(5), 725-743.
226
Bibliography
Kelle, P. & Silver, E. A. (1990b), 'Decreasing expected shortages through order splitting',
Engineering Costs & Production Economics 19(1-3), 351-357.
Kemahlioğlu-Ziya, E. (2004), 'Formal methods of value sharing in supply chains', PhD
thesis, School of Industrial and Systems Engineering, Georgia Institute of Technology,
Atlanta, GA, July 2004, http://smartech.gatech.edu/bitstream/1853/4965/1/kemahlioglu
ziya_eda_200407_phd.pdf, accessed 09/10/2010.
Khazanchi, D. (2005), 'Information technology (IT) appropriateness: The contingency
theory of "fit" and IT implementation in small and medium enterprises', Journal of
Computer Information Systems 45(3), 88-95.
Kibat, K.-D. (2007), 'Steigende Kosten, Überkapazitäten und Preisdruck: Das für die
deutsche Papierindustrie wichtige Inlandsgeschäft entwickelt sich sehr positiv', Holz-
Zentralblatt 51/52(Zellstoff- und Papierindustrie), 1467.
Kieser, A. & Kubicek, H. (1992), Organisation, De Gruyter.
Kim, J.-S. & Benjaafar, S. (2002), 'On the benefits of inventory-pooling in production-
inventory systems', Manufacturing & Service Operations Management 4(1), 12-16.
Kim, K. & Chhajed, D. (2000), 'Commonality in product design: Cost saving, valuation
change and cannibalization', European Journal of Operational Research 125(3), 602-
621.
Kim, K. & Chhajed, D. (2001), 'An experimental investigation of valuation change due to
commonality in vertical product line extension', Journal of Product Innovation
Management 18(4), 219-230.
Kisiel, R. (2000), 'Minneapolis test may shape GM's national web strategy', Automotive
News 75(5904), 1-2.
Kisperska-Moroñ, D. (2003), 'Responsibilities for inventory decisions in Polish
manufacturing companies', International Journal of Production Economics 81-82(1),
129-139.
Kistner, K.-P. & Steven, M. (2002), Betriebswirtschaftslehre im Grundstudium 1:
Produktion, Absatz, Finanzierung, Physica Verlag.
Klaas, T. (2002), Logistik-Organisation. Ein konfigurationstheoretischer Ansatz zur
logistikorientierten Organisationsgestaltung, DUV.
Klijn, F. & Slikker, M. (2005), 'Distribution center consolidation games', Operations
Research Letters 33(3), 285-288.
Klose, A. & Stähly, P. (2000), Standortwahl, Modelle und Methoden, in Klaus, P. &
Krieger, W., eds., Gabler Lexikon Logistik: Management logistischer Netzwerke und
Flüsse, Gabler, pp. 434-440.
Klose, A. (2001), Standortplanung in distributiven Systemen: Modelle, Methoden,
Anwendungen, Physica.
Kloster, T. (2002), Gestaltung von Logistik auf Basis von Netzeffekten, Lang.
Knight, F. H. (2005), Risk, Uncertainty and Profit, Cosimo, originally published by the
University of Chicago Press in 1921.
Köchel, P. (1975), 'A stochastic inventory model for some interconnected locations',
Mathematische Operationsforschung und Statistik, Series Optimization 6(3), 413-426.
227
Bibliography
Köchel, P. (1977), 'About the optimal inventory control in a system of locations: An

approximate solution', Mathematische Operationsforschung und Statistik, Series Opti-
mization 8(1), 105-118.
Köchel, P. (1982), 'A dynamic multi-location inventory model with transshipments
between locations', Mathematische Operationsforschung und Statistik, Series Optimiza-
tion 13(2), 267-286.
Köchel, P. (1988), 'Optimal adaptive inventory control for a multi-location model with
redistribution', Mathematische Operationsforschung und Statistik, Series Optimization
19(4), 525-537.
Köchel, P. (1998a), A survey on multi-location inventory models with lateral trans-
shipments, in Ganas, I. & Papachristos, S., eds., 'Proceedings of the Third ISIR Summer
School on Inventory Modelling', University of Ioannina, Department of Mathematics,
pp. 183-207.
Köchel, P. (1998b), Retrospective optimization of a two-location inventory model with
lateral transshipments, in 'Proceedings of the 2nd International Conference on Traffic
Science', pp. 129-139.
Köchel, P. (2007), 'Order optimisation in multi-location models with hub-and-spoke
structure', International Journal of Production Economics 108(1-2), 368-387.
Kohn, C. & Brodin, M. H. (2008), 'Centralised distribution systems and the environment:
How increased transport work can decrease the environmental impact of logistics',
International Journal of Logistics Research and Applications 11(3), 229-245.
Kohn, C. (2005), 'Centralisation of distribution systems and its environmental effects', PhD
thesis, Linköpings Universitet, Department of Management and Economics, Logistics
Management, May 2005, http://www.diva-portal.org/smash/get/diva2:20333/FULLTEX
T01, accessed 09/10/2010.
Köhne, M. F. (2007), Risikoartenübergreifende Steuerung in Industrieunternehmen in
Kaiser, T., ed., Wettbewerbsvorteil Risikomanagement: Erfolgreiche Steuerung der
Strategie-, Reputations- und operationellen Risiken, Erich Schmidt Verlag, pp. 307-326.
Kopczak, L. & Lee, H. (1994), Hewlett-Packard: Desktop printer supply chain (A), (B),
Stanford University case, Stanford Graduate School of Business.
Korovessi, E. & Linninger, A. A. (2006), Batch Processes, CRC Press.
Kotha, S. (1995), 'Mass customization: Implementing the emerging paradigm for
competitive advantage', Strategic Management Journal 16(1), 21-42.
Kotzab, H. (2000), Zum Wesen von Supply Chain Management vor dem Hintergrund der
betriebswirtschaftlichen Logistikkonzeption - erweiterte Überlegungen, in Wildemann,
H., ed., Supply Chain Management, TCW Transfer-Centrum-Verlag, pp. 21-47.
Kranenburg, A. A. & Van Houtum, G. J. (2009), 'A new partial pooling structure for spare
parts networks', European Journal of Operational Research 199(3), 908-921.
Kreng, V. B. & Lee, T. P. (2004), 'Modular product design with grouping genetic
algorithm - a case study', Computers & Industrial Engineering 46(3), 443-460.
Krishnan, K. S. & Rao, V. R. K. (1965), 'Inventory control in N warehouses', Journal of
Industrial Engineering 16(3), 212-215.
Krishnan, V. & Gupta, S. (2001), 'Appropriateness and impact of platform-based product
development', Management Science 47(1), 52-68.
228
Bibliography
Kroll, K. (2006), 'Virtual inventories get real', Multichannel Merchant 2(6), 1-54.
Kuhlang, P. & Matyas, K. (2005), 2 Entwicklung von Logistikprozessen in Schäppi, B.;
Andreasen, M. M.; Kirchgeorg, M. & Radermacher, F. J., eds., Handbuch
Produktentwicklung, Carl Hanser Verlag, pp. 657-676.
Kukreja, A. & Schmidt, C. P. (2005), 'A model for lumpy demand parts in a multi-location
inventory system with transshipments', Computers & Operations Research 32(8), 2059-
2075.
Kukreja, A.; Schmidt, C. P. & Miller, D. M. (2001), 'Stocking decisions for low-usage
items in a multilocation inventory system', Management Science 47(10), 1371-1383.
Kulkarni, S. S.; Magazine, M. J. & Raturi, A. S. (2005), 'On the trade-offs between risk
pooling and logistics costs in a multi-plant network with commonality', IIE
Kulpa, J. (2001), 'Magazine distributors pursue partnerships', Drug Store News 23(6), 33.
Kumar, A. & Schwarz, L. B. (1995), 'Risk-pooling along a fixed delivery route using a
dynamic inventory-allocation policy', Management Science 41(2), 344-362.
Kumar, S. & Craig, S. (2007), 'Dell, Inc.'s closed loop supply chain for computer assembly
plants', Information Knowledge Systems Management 6(3), 197-214.
Kumar, S. & Wilson, J. (2009), 'A manufacturing decision framework for minimizing
inventory costs of a configurable off-shored product using postponement', International
Journal of Production Research 47(1), 143-162.
Kumar, S.; Nottestad, D. A. & Murphy, E. E. (2009), 'Effects of product postponement on
the distribution network: A case study', Journal of the Operational Research Society
60(4), 471-480.
Kutanoglu, E. & Mahajan, M. (2009), 'An inventory sharing and allocation method for a
multi-location service parts logistics network with time-based service levels', European
Journal of Operational Research 194(3), 728-742.
Kutanoglu, E. (2008), 'Insights into inventory sharing in service parts logistics systems
with time-based service levels', Computers & Industrial Engineering 54(3), 341-358.
Labro, E. (2004), 'The cost effects of component commonality: A literature review through
a management-accounting lens', Manufacturing & Service Operations Management
6(4), 358-367.
Lahey, A. (1997), 'Less is more', Marketing Magazine 102(35), 12.
Larson, P. D. & Halldórsson, Á. (2004), 'Logistics versus supply chain management: An
international survey', International Journal of Logistics: Research and Applications
7(1), 17-31.
Lau, H. S. & Lau, A. H. (1994), 'Coordinating two suppliers with offsetting lead time and
price performance', Journal of Operations Management 11(4), 327-337.
Lau, H. S. & Zhao, L. G. (1993), 'Optimal ordering policies with two suppliers when lead
times and demands are all stochastic', European Journal of Operational Research 68(1),
120-133.
Lau, H. S. & Zhao, L. G. (1994), 'Dual sourcing cost-optimization with unrestricted lead-
time distributions and order-split proportions', IIE Transactions 26(5), 66-75.
229
Bibliography
Lau, R. S. M. (1995), 'Mass customization: The next industrial revolution', Industrial

Laudon, K. C. & Laudon, J. P. (2010), Management Information Systems: Managing the
Digital Firms, Prentice Hall.
Lawrence, K. D. & Zanakis, S. H. (1984), Production Planning and Scheduling:
Mathematical Programming Applications, Industrial Engineering and Management
Press, Institute of Industrial Engineers.
LeBlanc, L. J.; Hill, J. A.; Harder, J. & Greenwell, G. W. (2009), 'Modeling uncertain
forecast accuracy in supply chains with postponement', Journal of Business Logistics
30(1), 19-31.
Lee, D.-J. & Jeong, I.-J. (2009), 'Regression approximation for a partially centralized
inventory system considering transportation costs', Computers and Industrial
Engineering 56(4), 1169-1176.
Lee, H. L. & Nahmias, S. (1993), Chapter 1: Single-product, single-location models, in
Graves, S. C.; Rinnooy Kan, A. H. G. & Zipkin, P. H., eds., Logistics of Production and
Inventory, Elsevier, pp. 3-56.
Lee, H. L. & Tang, C. S. (1997), 'Modelling the costs and benefits of delayed product
differentiation', Management Science 43(1), 40-53.
Lee, H. L. & Tang, C. S. (1998), 'Variability reduction through operations reversal',
Lee, H. L. (1987), 'A multi-echelon inventory model for repairable items with emergency
lateral transshipments', Management Science 33(10), 1302-1316.
Lee, H. L. (1996), 'Effective inventory and service management through product and
process redesign', Operations Research 44(1), 151-159.
Lee, H. L. (1998), Postponement for mass customization: Satisfying customer demands for
tailor-made products, in Gattorna, J., ed., Strategic Supply Chain Alignment, Gower, pp.
77-91.
Lee, H. L.; Billington, C. & Carter, B. (1993), 'Hewlett Packard gains control of inventory
and service through design for localization', Interfaces 23(4), 1-11.
Lee, H. L.; Padmanabhan, V. & Whang, S. (1997a), 'The bullwhip effect in supply chains',
Sloan Management Review 38(3), 93-102.
Lee, H. L.; Padmanabhan, V. & Whang, S. (1997b), 'Information distortion in a supply
chain: The bullwhip effect', Management Science 43(4), 546-558.
Lee, H. L.; Padmanabhan, V. & Whang, S. (2004), 'Information distortion in a supply
chain: The bullwhip effect', Management Science 50, 1875-1886.
Lee, Y. H.; Jung, J. W. & Jeon, Y. S. (2007), 'An effective lateral transshipment policy to
improve service level in the supply chain', International Journal of Production
Economics 106(1), 115-126.
Lehrer, E. (2002), 'Allocation processes in cooperative games', International Journal of
Game Theory 31, 341-351.
Lemoine, O. W. & Skjoett-Larsen, T. (2004), 'Reconfiguration of supply chains and
implications for transport: A Danish study', International Journal of Physical
230
Bibliography
Li, J.; Cheng, T. C. E. & Wang, S. (2007), 'Analysis of postponement strategy for
perishable items by EOQ-based models', International Journal of Production
Economics 107(1), 31-38.
Li, J.; Wang, S. & Cheng, T. E. (2008), 'Analysis of postponement strategy by EPQ-based
models with planned backorders', Omega 36(5), 777-788.
Li, L. (2007), Supply Chain Management: Concepts, Techniques and Practices Enhancing
Value through Collaboration, World Scientific.
Li, Q.; Wu, X. & Cheung, K. L. (2009), 'Optimal policies for inventory systems with
separate delivery-request and order-quantity decisions', Operations Research 57(3),
626-636.
Lim, W.-S.; Ou, J. & Teo, C.-P. (2002), 'Inventory cost effect of consolidating several one-
warehouse multiretailer systems', Operations Research 51(4), 668-672.
Lin, C.-T.; Chen, C.-B. & Hsieh, H.-J. (2001a), 'Effects of centralization on expected
profits in a multi-location newsboy problem', Journal of the Operational Research
Society 52(7), 839-844.
Lin, G. Y.; Breitwieser, R.; Cheng, F.; Eagen, J. T. & Ettl, M. (2001b), Product hardware
complexity and its impact on inventory and customer on-time delivery (industrial
application case: IBM), in Shaw, M. J., ed., Information-Based Manufacturing:
Technology, Strategy, and Applications, Kluwer, pp. 61-80.
Liu, J. & Lee, C.-G. (2007), 'Evaluation of inventory policies with unidirectional
substitutions', European Journal of Operational Research 182(1), 145-163.
Logistics & Transport Focus (2001), 'Living better with public transport', Logistics &
Transport Focus 3(1), 14.
Lovell, A.; Saw, R. & Vermeulen, P. (2005), 'Sony's vision: Sales up 80% inventory down
80%', Logistics & Transport Focus 7(5), 40-44.
Lowe, D. (2002), The Dictionary of Transport and Logistics, Kogan Page Publishers.
Lue, N.-H. (2006), 'The simulation analysis of lateral transshipment and risk pooling for
petrochemical supply chains', Master's thesis, Department of Logistics Management,
National Kaohsiung First University of Science and Technology, Yenchao, Kaohsiung,
Taiwan, Republic of China, 10/26/2006, http://ethesys.nkfust.edu.tw/ETD-db/ETD-
search/view_etd?URN=etd-1026106-165702, accessed 03/29/2008.
Lüpschen, B. (2004), 'Kostendegressionspotenziale in Logistiksystemen', technical report,
Arbeitsbericht 105, University of Cologne, Department of Business Policy and
Logistics, http://www.spl.uni-koeln.de/fileadmin/user/dokumente/forschung/arbeitsbe
richte/SPL-Arbeitsbericht_105-Luepschen.pdf, accessed 04/27/2008.
Ma, S.; Wang, W. & Liu, L. (2002), 'Commonality and postponement in multistage
assembly systems', European Journal of Operational Research 142(3), 523-538.
Magretta, J. (1998), 'The power of virtual integration: An interview with Dell Computer's
Michael Dell', Harvard Business Review 76(2), 72-84.
Mahajan, S. & Van Ryzin, G. (2001a), 'Stocking retail assortments under dynamic
consumer substitution', Operations Research 49(3), 334-351.
Mahajan, S. & Van Ryzin, G. (2001b), 'Inventory competition under dynamic consumer
choice', Operations Research 49(5), 646-657.
231
Bibliography
Mahar, S. & Wright, P. D. (2009), 'The value of postponing online fulfillment decisions in
multi-channel retail/e-tail organizations', Computers & Operations Research 36(11),
3061-3072.
Mahar, S.; Bretthauer, K. M. & Venkataramanan, M. A. (2009), 'The value of virtual
pooling in dual sales channel supply chains', European Journal of Operational
Research 192(2), 561-575.
Mahmoud, M. M. (1992), 'Optimal inventory consolidation schemes: A portfolio effects
analysis', Journal of Business Logistics 13(1), 193-214.
Maister, D. H. (1976), 'Centralisation of inventories and the "square root law"',
International Journal of Physical Distribution 6(3), 124-134.
Makridakis, S.; Andersen, A.; Carbone, R.; Filder, R.; Hibon, M.; Lewandowski, R.;
Newton, J.; Parzen, E. & Winkler, R. (1982), 'The accuracy of extrapolation (time
series) methods: Results of a forecasting competition', Journal of Forecasting 1(2), 111-
153.
March, C. (2009), Business Organisation for Construction, Taylor & Francis.
Martinez, S. L.; Arostegui, M. A. & Brady, S. P. (2002), 'Logistics support: Improving the
logistics pipeline: Achieving agile combat supply support', Air Force Journal of
Logistics 26(4), 12-21, 45.
Maskell, B. (1991), Performance Measurement for World Class Manufacturing: A Model
for American Companies, Productivity Press.
Mason, J.; Scott, P.; Ribera, M.; Farris, J. A. & Kirk, R. G. (2003), 'Integrating the
warehousing and the transportation functions of the supply chains', Transportation
Research Part E 39(2), 141-159.
Masters, J. M. (1980), 'The effect of freight consolidation on consumer service', Journal of
Mathaisel, D. F. X.; Manary, J. M. & Comm, C. L. (2009), Military Enterprise
Sustainability: Enhancing the Military's Ability to Perform its Mission, CRC Press
Taylor & Francis Group.
Matthews, P. & Syed, N. (2004), 'The power of postponement', Supply Chain Management
Review 8(3), 28-34.
Mayne, E.; Winter, D. & Schweinsberg, C. (2008), 'Production puzzle', Ward's AutoWorld,
08/01/2008, http://wardsautoworld.com/ar/auto_production_puzzle/index.html, access
09/04/2008.
McClain, J. O.; Maxwell, W. L.; Muckstadt, J. A.; Thomas, L. J. & Weiss, E. N. (1984),
'Comment on "Aggregate safety stock levels and component part commonality"',
McDermott, C. M. & Stock, G. N. (1994), 'The use of common parts and designs in high-
tech industries: A strategic approach', Production & Inventory Management Journal
35(3), 65-69.
McGavin, E. J.; Schwarz, L. B. & Ward, J. E. (1993), 'Two-interval inventory-allocation
policies in a one-warehouse n-identical-retailer distribution system', Management
Science 39(9), 1092-1107.
McGillivary, A. & Silver, E. (1978), 'Some concepts for inventory control under
substitutable demand', INFOR 16(1), 47-63.
232
Bibliography
McKinnon, A. & Forster, M. (2000), 'Full report of the delphi 2005 survey European
logistical and supply chain trends: 1999-2005', technical report, Heriot-Watt University,
Edinburgh, UK, http://www.sml.hw.ac.uk/logistics/pdf/delphi.pdf, accessed 06/03/2008.
McKinnon, A. (1997), Case 16 - Scottish brewers: The restructuring of a depot system, in
Taylor, D. H., ed., Global Cases in Logistics and Supply Chain Management: Teachers'
Manual, International Thomson Business Press, pp. 78-83.
McKinnon, A. C. (1989), Physical Distribution Systems, Routledge.
McKinnon, A. C. (2003), 'The effects of transport investment on logistical efficiency',
Heriot-Watt University, Logistics Research Centre, Edinburgh, UK, 03/20/2003,
http://www.sml.hw.ac.uk/logistics/downloads/congestion-reliabilitystudy/Transport%20
Investment%20and%20Logistical%20Efficiency%20(McKinnon%20-%20TSU%20Ox
ford).pdf, accessed 09/09/2008.
Memon, F. (1997), 'Virtual warehouse keeps Toyota running', Inter@ctive Week 3(17), 40.
Mercer, A. & Tao, X. (1996), 'Alternative inventory and distribution policies of a food
manufacturer', Journal of the Operational Research Society 47(6), 755-765.
Meyer, A. D.; Tsui, A. S. & Hinings, C. R. (1993), 'Configurational approaches to
organizational analysis', Academy of Management Journal 36(6), 1175-1195.
Meyer, M. H. & Lehnerd, A. P. (1997), The Power of Product Platforms: Building Value
and Cost Leadership, The Free Press.
Meyr, H. (2003), 'Die Bedeutung von Entkopplungspunkten für die operative Planung von
Supply Chains', Zeitschrift für Betriebswirtschaft 73(9), 941-962.
Miller, D. & Friesen, P. H. (1978), 'Archetypes of strategy formulation', Management
Science 24(9), 921-933.
Miller, D. & Friesen, P. H. (1980), 'Archetypes of organizational transition', Administrative
Science Quarterly 25(2), 268-299.
Miller, D. (1981), 'Toward a new contingency approach: The search for organizational
gestalts', Journal of Management Studies 18(1), 1-26.
Milner, J. M. & Kouvelis, P. (2007), 'Inventory, speculation, and sourcing strategies in the
presence of online exchanges', Manufacturing & Service Operations Management 9(3),
312-331.
Minkowski, H. (1896), Geometrie der Zahlen, Teubner, reprint by VDM Verlag Dr.
Müller, 2006.
Minner, S. & Silver, E. A. (2005), 'Evaluation of two simple extreme transshipment
strategies', International Journal of Production Economics 93-94(1), 1-11.
Minner, S. (2003), 'Multiple-supplier inventory models in supply chain management: A
review', International Journal of Production Economics 81-82(1), 265-279.
Minner, S.; Silver, E. A. & Robb, D. J. (2003), 'An improved heuristic for deciding on
emergency transshipments', European Journal of Operational Research 148(2), 384-
400.
Miranda, P. A. & Garrido, R. A. (2004), 'Incorporating inventory control decisions into a
strategic distribution network design model with stochastic demand', Transportation
Research: Part E 40(3), 183-207.
233
Bibliography
Mirchandani, P. & Mishra, A. K. (2002), 'Component commonality: Models with product-

specific service constraints', Production & Operations Management 11(2), 199-215.
Mishra, A. K. & Tadikamalla, P. R. (2006), 'Order splitting in single sourcing with
scheduled-release orders', Journal of the Operational Research Society 57(2), 177-189.
MMR (2009), 'Impact brands should win in SKU rationalization', MMR 26(15), 19.
Modern Materials Handling (2002), 'Number of warehouses continues to shrink', Modern
Materials Handling 57(14), 1.
Mohebbi, E. & Choobineh, F. (2005), 'The impact of component commonality in an
assemble-to-order environment under supply and demand uncertainty', Omega 33(6),
472-482.
Mohebbi, E. & Posner, M. J. M. (1998), 'Sole versus dual sourcing in a continuous-review
inventory system with lost sales', Computers in Industrial Engineering 34(2), 321-336.
Moinzadeh, K. & Nahmias, S. (1988), 'A continuous review model for an inventory system
with two supply modes', Management Science 34(6), 761-773.
Monezka, R. M. & Carter, P. L. (1976), The economics of vendor pooling, in 'Academy of
Management Proceedings', pp. 207-211.
Monopolkommission (1986), Gesamtwirtschaftliche Chancen und Risiken wachsender
Unternehmensgrößen: Hauptgutachten 1984/1985, Nr. 6, Nomos Verlagsgesellschaft.
Mood, A. M.; Graybill, F. A. & Boes, D. C. (1974), Introduction to the Theory of
Statistics, McGraw-Hill.
Morehouse, J. E. & Bowersox, D. J. (1995), 'Supply chain management: Logistics for the
future', technical report, Food Marketing Institute, Washington, D.C.
Movianto GmbH (2010), 'Postponement', http://www.movianto.com/services-solutions/ma
nufacturing-postponement.htm, accessed 01/22/2010.
Moyer, R. C.; McGuigan, J. R. & J., K. W. (1992), Contemporary Financial Management,
West Publishing Company.
Muckstadt, J. A. & Thomas, L. J. (1980), 'Are multi-echelon inventory methods worth
implementing in systems with low demand rate items?', Management Science 26(5),
483-494.
Muckstadt, J. A. (2005), Analysis and Algorithms for Service Parts Supply Chains,
Springer.
Müller, A.; Scarsini, M. & Shaked, M. (2002), 'The newsvendor game has a nonempty
core', Games and Economic Behavior 38(1), 118-126.
Müller-Benedict, V. (2007), Grundkurs Statistik in den Sozialwissenschaften, VS Verlag
für Sozialwissenschaften.
Müller-Nedebock, S. (2009), Erfolgsfaktoren der Strategieimplementierung im Strategi-
schen Management: Forschungsstand und Entwicklungstendenzen, IGEL Verlag.
Naesens, K.; Gelders, L. & Pintelon, L. (2007), 'A swift response tool for measuring the
strategic fit for resource pooling: A case study', Management Decision 45(3), 434-449.
Nahmias, S. (2005), Production and Operations Analysis, McGraw-Hill/Irwin.
Nair, A. (2005), 'Linking manufacturing postponement, centralized distribution and value
chain flexibility with performance', International Journal of Production Research 43(3),
447-463.
234
Bibliography
Narus, J. A. & Anderson, J. C. (1996), 'Rethinking distribution: Adaptive channels',

Harvard Business Review 74(4), 112-120.
Naseraldin, H. & Herer, Y. T. (2008), 'Integrating the number and location of retail outlets
on a line with replenishment decisions', Management Science 54(9), 1666-1683.
Neale, J. J.; Tomlin, B. T. & Willems, S. P. (2003), The role of inventory in superior
supply chain performance, in Harrison, T. P.; Lee, H. L. & Neale, J. J., eds., The
Practice of Supply Chain Management: Where Theory and Application Converge,
Kluwer, pp. 31-60.
Needham, P. M. & Evers, P. T. (1998), 'The influence of individual cost factors on the use
of emergency transshipments', Transportation Research: Part E: Logistics and
Netessine, S. & Rudi, N. (2003), 'Centralized and competitive inventory models with
demand substitution', Operations Research 51(2), 329-335.
Netessine, S. & Rudi, N. (2006), 'Supply chain choice on the internet', Management
Science 52(6), 844-864.
Netessine, S. (2005), 'From theory to empirical evidence and back: Integrating data-driven
research into your portfolio', Purdue CIBER Consortium meeting on International
Operations Management, 09/17/2005.
Neumann, K. (1996), Produktions- und Operations-Management, Springer.
Neus, W. (2009), Einführung in die Betriebswirtschaftslehre aus institutionenökonomi-
scher Sicht, Mohr Siebeck.
Newson, P. L. (1978), The Future Role of Depots in a Distribution Network, Post Office.
Nobre, F. S.; Tobias, A. M. & Walker, D. S. (2009), Organizational and Technological
Implications of Cognitive Machines: Designing Future Information Management
Systems, Idea Group Inc. (IGI).
Nonås, L. M. & Jörnsten, K. (2005), Heuristics in the multi-location inventory system with
transshipments, in Kotzab, H.; Seuring, S.; Müller, M. & Reiner, G., eds., Research
Methodologies in Supply Chain Management, Physica Verlag, pp. 509-524.
Nonås, S. L. (2009), 'Finding and identifying optimal inventory levels for systems with
common components', European Journal of Operational Research 193(1), 98-119.
Nozick, L. K. & Turnquist, M. A. (2001), 'Inventory, transportation, service quality and the
location of distribution centers', European Journal of Operational Research 129(2),
362-371.
Obal, P. (2006), Glossary of Supply Chain Terminology: A Dictionary on Technology,
Logistics, Transportation, Warehousing, Manufacturing, Purchasing, and More!,
Verlag Industrial Data & Information Inc.
Olsson, F. (2009), 'Optimal policies for inventory systems with lateral transshipments',
Oracle Corporation (2004), 'The adaptive supply chain: Postponement for profitability',
http://www.oracle.com/us/solutions/scm/018543.pdf, accessed 09/08/2009.
Oracle Corporation (2009), 'JD Edwards OneWorld Manual', Redwood Shores, CA.
Orgel, D. (2009), 'How partnerships succeeding in time of friction', SN: Supermarket News
57(41), 12.
235
Bibliography
Orlicky, J. A. (1975), Material Requirements Planning, McGraw-Hill.

Osteryoung, J. S.; Nosari, E.; McCarty, D. E. & Reinhart, W. J. (1986), 'Use of the EOQ
model for inventory analysis', Production and Inventory Management 27(3), 39-46.
Özdemir, D.; Yücesan, E. & Herer, Y. T. (2006a), Multi-location transshipment problem
with capacitated production and lost sales, in 'Proceedings of the 2006 Winter
Simulation Conference', pp.1470-1476.
Özdemir, D.; Yücesan, E. & Herer, Y. T. (2006b), 'Multi-location transshipment problem
with capacitated transportation', European Journal of Operational Research 175(1),
602-621.
Özen, U. & Sošić, G. (2006), 'A multi-retailer decentralized distribution system with
updated demand information', BETA Working Paper 193, Eindhoven University of
Technology, 08/18/2006, http://cms.ieis.tue.nl/Beta/Files/WorkingPapers/Beta_WP193.
Özen, U.; Fransoo, J.; Norde, H. & Slikker, M. (2008), 'Cooperation between multiple
newsvendors with warehouses', Manufacturing & Service Operations Management
10(2), 311-324.
Özen, U.; Norde, H. & Slikker, M. (2005), 'On the convexity of newsvendor games',
Discussion Paper 103, Tilburg University, Center for Economic Research, 08/30/2005,
http://ssrn.com/abstract=642401, accessed 09/10/2010.
Özen, U.; Sošić, G. & Slikker, M. (2010), 'A collaborative decentralized distribution
system with demand forecast updates', 04/07/2010, http://ssrn.com/abstract=929652, ac-
cessed 09/08/2010.
Özer, Ö. & Xiong, H. (2008), 'Stock positioning and performance estimation for
distribution systems with service constraints', IIE Transactions 40(12), 1141-1157.
Özer, Ö. (2003), 'Replenishment strategies for distribution systems under advance demand
information', Management Science 49(3), 255-272.
Ozsen, L.; Coullard, C. R. & Daskin, M. S. (2008), 'Capacitated warehouse location model
with risk pooling', Naval Research Logistics 55(4), 295-312.
Pagh, J. D. & Cooper, M. C. (1998), 'Supply chain postponement and speculation
strategies: How to choose the right strategy', Journal of Business Logistics 19(2), 13-33.
Pamplin, C. (2002), 'Casting a cold eye', Convenience Store News 38(9), 32-33.
Pan, A. C. & Liao, C.-J. (1989), 'An inventory model under just-in-time purchasing
agreements', Production & Inventory Management Journal 30(1), 49-52.
Pan, A. C.; Ramasesh, R. V.; Hayya, J. C. & Ord, J. K. (1991), 'Multiple sourcing: The
determination of lead times', Operations Research Letters 10(1), 1-7.
Paraschis, I. N. (1989), Optimale Gestaltung von Mehrprodukt-Distributionssystemen:
Modelle - Methoden - Anwendungen, Physica Verlag.
Parlar, M. & Goyal, S. (1984), 'Optimal ordering decisions for two substitutable products
with stochastic demands', Opsearch 21(1), 1-15.
Pasin, F.; Jobin, M.-H. & Cordeau, J.-F. (2002), 'An application of simulation to analyse
resource sharing among health-care organisations', International Journal of Operations
& Production Management 22(4), 381-393.
236
Bibliography
Pasternack, B. A. & Drezner, Z. (1991), 'Optimal inventory policies for substitutable

commodities with stochastic demand', Naval Research Logistics 38, 221-240.
Paterson, C.; Kiesmüller, G.; Teunter, R. & Glazebrook, K. (2009), 'Inventory models with
lateral transshipments: A review', BETA Working Paper 287, Eindhoven University of
Technology, 08/26/2009, http://cms.ieis.tue.nl/Beta/Files/WorkingPapers/Beta_wp287.
Petering, M. E. H. & Murty, K. G. (2009), 'Effect of block length and yard crane
deployment systems on overall performance at a seaport container transshipment
terminal', Computers & Operations Research 36(5), 1711-1725.
Peterson, R. (1969), 'A note on the determination of optimal forecasting strategy',
Management Science 16(4), B165-B169.
Pfeifer, B. (2010), Zur Nachhaltigkeitsorientierung von Private Equity-Investoren, Josef
Eul Verlag GmbH.
Pfohl, H.-C. (1994), Logistikmanagement: Funktionen und Instrumente, Springer.
Pfohl, H.-C. (2004a), Logistiksysteme: Betriebswirtschaftliche Grundlagen, Springer.
Pfohl, H.-C. (2004b), Logistikmanagement: Konzeption und Funktionen, Springer.
Pfohl, H.-C.; Zöllner, W. A. & Weber, N. (1992), 'Economies of scale in customer
warehouses: Theoretical and empirical analysis', Journal of Business Logistics 13(1),
95-124.
Pine II, B. J.; Victor, B. & Boynton, A. C. (1993), 'Making mass customization work',
Harvard Business Review 71(5), 108-118.
Pinto, D. (2009a), 'Outsiders bring new thinking to industry', Chain Drug Review 31(15),
16.
Pinto, D. (2009b), 'A merchandising cataclysm', MMR 26(15), 8.
Piontek, J. (2007), Bausteine des Logistikmanagements: Supply-Chain-Management, E-
Logistics, Logistikcontrolling, Verlag Neue Wirtschafts-Briefe GmbH & Co.
Pishchulov, G. & Richter, K. (2009), 'Inventory rationing and sharing in pre-sell
distribution with mobile communication technologies', International Journal of
Production Economics 121(2), 584-600.
Pishchulov, G. (2008), Inventory Rationing and Sharing in Pre-Sell Distribution with
Mobile Communication Technologies, Shaker.
Plambeck, E. & Taylor, T. (2003), 'Sell the plant? The impact of contract manufacturing on
innovation, capacity, and profitability', technical report, Graduate School of Business,
Stanford University, Palo Alto, Management Science, 51(1), 2005, 133-150.
Planning & Reporting (2001), 'How Insight Enterprises benchmarks 'virtual' inventory
turnover', Financial Analysis, Planning & Reporting 1(5), 1-5.
Porter, M. E. (1985), Competitive Advantage, Free Press.
Porter, M. E. (2008), On Competition, Harvard Business School Publishing.
Pringle, D. (2003), 'Nokia eschews factories in most low-cost regions', The Wall Street
Journal, 01/03/2003.
Progress Software Corporation (2007), 'Improving the business model with Progress Open-
Edge®', http://www.progress.com/progress_software/docs/kingfield_health.pdf, access
06/02/2008.
237
Bibliography
Pugh, D. S. (1981), The Aston programme perspective, in Van de Ven, A. H. & Joyce, W.
F., eds., Perspectives on Organization Design and Behavior, Wiley, pp. 135-166.
Purchasing (2000), 'Wareforce.com joins group of Commerce One suppliers', Purchasing
128(8), 132.
Qi, X. (2007), 'Order splitting with multiple capacitated suppliers', European Journal of
Rabinovich, E. & Evers, P. T. (2003a), 'Product fulfillment in supply chains supporting
internet-retailing operations', Journal of Business Logistics 24(2), 205-236.
Rabinovich, E. & Evers, P. T. (2003b), 'Postponement effects on inventory performance
and the impact of information systems', International Journal of Logistics Management
14(1), 33-48.
Raghunathan, S. (2003), 'Impact of demand correlation on the value of and incentives for
information sharing in a supply chain', European Journal of Operational Research
146(3), 634-649.
Rajaram, K. & Tang, C. (2001), 'The impact of product substitution on retail
merchandising', European Journal of Operational Research 135(3), 582-601.
Ramasesh, R. V. (1990), 'Recasting the traditional inventory model to implement just-in-
time purchasing', Production and Inventory Management 31(1), 71-75.
Ramasesh, R. V. (1991), 'Procurement under uncertain supply lead times–a dual-sourcing
technique could save costs', Engineering Costs and Production Economics 21(1), 59-68.
Ramasesh, R. V.; Ord, J. K. & Hayya, J. C. (1993), 'Note: Dual sourcing with nonidentical
suppliers', Naval Research Logistics 40(2), 279-288.
Ramasesh, R. V.; Ord, J. K.; Hayya, J. C. & Pan, A. (1991), 'Sole versus dual sourcing in
stochastic lead-time (s, Q) inventory models', Management Science 37(4), 428-443.
Ramdas, K. & Sawhney, M. S. (2001), 'A cross-functional approach to evaluating multiple
line extensions for assembled products', Management Science 47(1), 22-36.
Randall, T.; Netessine, S. & Rudi, N. (2002), 'Should you take the virtual fulfillment
path?', Supply Chain Management Review 6(6), 54-58.
Randall, T.; Netessine, S. & Rudi, N. (2006), 'An empirical examination of the decision to
invest in fulfillment capabilities: A study of internet retailers', Management Science
52(4), 567-580.
Reihlen, M. (1997), Entwicklungsfähige Planungssysteme. Grundlagen, Konzepte und
Anwendungen zur Bewältigung von Innovationsproblemen, Gabler.
Reiner, G. (2005), Supply chain management research methodology using quantitative
models based on empirical data, in Kotzab, H.; Seuring, S.; Müller, M. & Reiner, G.,
eds., Research Methodologies in Supply Chain Management, Physica Verlag, pp. 431-
444.
Reyes, P. M. & Meade, L. M. (2006), 'Improving reverse supply chain operational
performance: A transshipment application study for not-for-profit organizations',
Journal of Supply Chain Management: A Global Review of Purchasing & Supply 42(1),
38-48.
Reynolds, H. T. (1984), Analysis of Nominal Data, Vol. 7, Sage.
238
Bibliography
Richter, K. & Dobos, I. (1999), 'Analysis of the EOQ repair and waste disposal problem
with integer setup numbers', International Journal of Production Economics 59(1), 463-
467.
Richter, K. & Dobos, I. (2003a), Production-inventory control in an EOQ-type reverse
logistics system, in Dyckhoff, H.; Lackes, R. & Reese, J., eds., Supply Chain
Management and Reverse Logistics, Springer, pp. 139-160.
Richter, K. & Dobos, I. (2003b), A reverse logistics model with integer setup numbers, in
Leopold-Wildburger, U.; Rendl, F. & Wäscher, G., eds., 'Operations Research
Proceedings 2002', Springer, pp. 95-101.
Richter, K. & Gobsch, B. (2005), 'Kreislauf-Logistik mit Losgrößenrestriktionen', ZfB
Special Issue 4/2005: Reverse Logistics II, 57-79.
Richter, K. & Sombrutzki, M. (2000), 'Remanufacturing planning for the reverse
Wagner/Whitin models', European Journal of Operational Research 121(2), 304-315.
Richter, K. & Weber, J. (2001), 'The reverse Wagner/Whitin model with variable
manufacturing and remanufacturing cost', International Journal of Production
Economics 71(1-3), 447-456.
Richter, K. (1996a), 'The extended EOQ repair and waste disposal model', International
Journal of Production Economics 45(1-3), 443-447.
Richter, K. (1996b), 'The EOQ repair and waste disposal model with variable setup
numbers', European Journal of Operational Research 95(2), 313-324.
Richter, K. (1997), 'Pure and mixed strategies for the EOQ repair and waste disposal
problem', OR-Spektrum 19(2), 123-129.
Robinson, L. W. (1990), 'Optimal and approximate policies in multiperiod, multilocation
inventory models with transshipments', Operations Research 38(2), 278-295.
Robinson, L. W. (1993), 'A comment on Gerchak and Gupta's "On apportioning cost to
customers in centralized continuous review inventory systems"', Journal of Operations
Romano, P. (2006), 'Supply chain risk pooling', Blackwell Encyclopedic Dictionary of
Operations Management, p. 320, http://www.blackwellreference.com/public/tocnode?id
=g9780631233176_chunk_g978140511096923_ss32-1, accessed 09/01/2010.
Ronen, D. (1990), 'Inventory centralization/decentralization–the "square root law" revisited
again', Journal of Business Logistics 11(2), 129-138.
Roque, I. M. (1977), Production-inventory system economy using a component
standardization factor, in 'Proceedings of the Midwest American Institute for Decision
Sciences'.
Ross, D. F. (1996), Distribution Planning and Control, Chapman & Hall.
Röttgen, W. (1987), 'Der Feinpapiergroßhandel als Absatzmittler für graphische Papiere',
Master's thesis, Universität zu Köln, Seminar für Handelsbetriebslehre, Prof. Dr. F.
Klein-Blenkers.
Rudi, N.; Kapur, S. & Pyke, D. F. (2001), 'A two-location inventory model with
transshipment and local decision making', Management Science 47(12), 1668-1680.
Rumyantsev, S. & Netessine, S. (2007), 'What can be learned from classical inventory
models? A cross-industry exploratory investigation', Manufacturing & Service
Operations Management 9(4), 409-429.
239
Bibliography
Ruusunen, J.; Ehtamo, H. & Hamalainen, R. (1991), 'Dynamic cooperative electricity

exchange in a power pool', IEEE Transactions on Systems, Management, and
Cybernetics 21(4), 758-766.
Ryan, T. (2009), 'Retailwire discussion: SKU rationalization reshapes retail shelves',
Private Label Buyer 23(9), 10.
Ryu, S. W. & Lee, K. K. (2003), 'A stochastic inventory model of dual sourced supply
chain with lead-time reduction', International Journal of Production Economics 81-
82(1), 513-527.
Sajadieh, M. S. & Eshghi, K. (2009), 'Sole versus dual sourcing under order dependent
lead times and prices', Computers & Operations Research 36(12), 3272-3280.
Samuelson, P. A. & Nordhaus, W. D. (1998), Volkswirtschaftslehre, Ueberreuter.
Savage, S.; Scholtes, S. & Zweidler, D. (2006), 'Probability management', OR/MS Today
33(1), 21-28.
Sawyers, A. (2003), 'Web E.L.V.I.S.: A whole lot of selling going on', Automotive News
77(6029), 6.
Scherer, A. G. & Beyer, R. (1998), 'Der Konfigurationsansatz im strategsichen
Management – Rekonstruktion und Kritik', Die Betriebswirtschaft (DBW) 58(3), 332-
347.
Scherer, F. M. & Ross, D. (1990), Industrial Market Structure and Economic
Performance, Houghton Mifflin Company.
Schmitt, A. J.; Snyder, L. V. & Shen, Z.-J. M. (2008a), 'Centralization versus
decentralization: Risk pooling, risk diversification, and supply uncertainty in a one-
warehouse multiple-retailer system', 02/29/2008, http://ssrn.com/abstract=1115392, ac-
cessed 06/02/2008.
Schmitt, A. J.; Snyder, L. V. & Shen, Z.-J. M. (2008b), Centralization versus
decentralization: Risk pooling, risk diversification, and supply uncertainty in a one-
warehouse multiple-retailer system, in 'Manufacturing & Service Operations
Management Conference Proceedings', June 5-6, 2008, http://www.lehigh.edu/~ajs404/
MSOM_conf_08_OWMR.pdf, accessed 06/02/2008.
Schneekloth, U. & Leven, I. (2003), 'Woran bemisst sich eine gute Bevölkerungsumfrage?
Analysen zu Ausmaß, Bedeutung und zu den Hintergründen von Nonresponse in
stichprobenbasierten Zufallserhebungen am Beispiel des ALLBUS', ZUMA-Nachrichten
53, 16-57.
Schneider, D. (1987), Allgemeine Betriebswirtschaftslehre, Oldenbourg.
Schnuckel, M. (2010), Optimierung der Beschaffung durch vertikale Kooperation: Zur
Relevanz des Bullwhip-Effekts aus der Perspektive des Einzelhandels, in Fröhlich-
Glantschnig, E. & Lingohr, T., eds., Gibt es die optimale Einkaufsorganisation?
Organisatorischer Wandel und pragmatische Methoden zur Effizienzsteigerung, Gabler,
pp. 147-166.
Schoenmeyr, T. (2005), '"Centralized ordering policies in a multi-warehouse system with
lead times and random demand" A paper by Gary Eppen and Linus Schrage: Presenta-
tion by Tor Schoenmeyr', Lecture Notes, Massachusetts Institute of Technology, http://
ocw.mit.edu/courses/sloan-school-of-management/15-764-the-theory-of-operations-ma
nagement-spring-2004/lecture-notes/lec2_schoenmeyr.pdf, accessed 09/10/2010.
240
Bibliography
Schreyögg, G. (1995), Umwelt, Technologie und Organisationsstruktur: Eine Analyse des

kontingenztheoretischen Ansatzes, Haupt.
Schreyögg, G. (2008), Organisation, Gabler.
Schröder, R. (2008), Strategische Orientierungen für junge Technologieunternehmen,
Gabler.
Schulte, C. (1999), Logistik: Wege zur Optimierung des Material- und Informationsflusses,
Vahlen.
Schulze, P. M. (2007), Beschreibende Statistik, Oldenbourg Wissenschaftsverlag.
Schwarz, L. B. (1981), 'Physical distribution: The analysis of inventory and location', AIIE
Schwarz, L. B. (1989), 'A model for assessing the value of warehouse risk-pooling: Risk-
pooling over outside-supplier leadtimes', Management Science 35(7), 828-842.
Schwarz, L. B.; Deuermeyer, B. L. & Badinelli, R. D. (1985), 'Fill-rate optimization in a
one-warehouse n-identical retailer distribution system', Management Science 31(4),
488-498.
Scott, W. R. (1981), Organizations: Rational, Natural, and Open Systems, Prentice Hall.
Sculli, D. & Shum, Y. W. (1990), 'Analysis of a continuous review stock-control model
with multiple suppliers', Journal of the Operational Research Society 41(9), 873-877.
Sculli, D. & Wu, S. Y. (1981), 'Stock control with two suppliers and normal lead times',
Journal of the Operational Research Society 32(11), 1003-1009.
SDM (2001), 'Virtual warehouse created for Edge members', SDM: Security Distributing
& Marketing 31(14), 26.
Sedarage, D.; Fujiwara, O. & Luong, H. T. (1999), 'Determining optimal order splitting
and reorder level for N-supplier inventory systems', European Journal of Operational
Research 116(2), 389-404.
Segal-Horn, S. & Faulkner, D. (2010), Understanding Global Strategy, South-Western
Cengage Learning.
Shah, J. (2009), Supply Chain Management: Text and Cases, Pearson.
Shao, J.; Krishnan, H. & McCormick, S. T. (2009), 'Incentives for transshipment in a
supply chain with decentralized retailers', technical report, Sauder School of Business,
University of British Columbia, Vancouver, 09/25/2009, http://people.commerce.ubc.ca
/phd/shao/Transshipment.pdf, accessed 01/07/2010.
Shao, X. & Ji, J. (2006), 'Reconfiguration of pharmaceutical logistics operations in China:
An empirical study', Transportation Journal 45(4), 52-66.
Shao, X.-F. & Ji, J.-H. (2008), 'Evaluation of postponement strategies in mass
customization with service guarantees', International Journal of Production Research
46(1), 153-171.
Shapiro, R. D. & Heskett, J. L. (1985), Logistics Strategy: Cases and Concepts, West
Publishing.
Shapiro, R. D. (1984), 'Get leverage from logistics', Harvard Business Review 62(3), 119-
126.
241
Bibliography
Shapley, L. S. (1953), A value for n-person games in Kuhn, H. W. & Tucker, A. W., eds.,
Contributions to the Theory of Games II (Annals of Mathematics Studies 28), Princeton
University Press, pp. 307-317.
Sheffi, Y. (2004), Demand variability and supply chain flexibility: Driving from pure cost
to lean flexibility supply chains, in Prockl, G.; Bauer, A.; Pflaum, A. & Müller-
Steinfahrt, U., eds., Entwicklungspfade und Meilensteine moderner Logistik: Skizzen
einer Roadmap, Gabler, pp. 85-117.
Sheffi, Y. (2006), Worst-Case-Szenario: Wie Sie Ihr Unternehmen auf Krisen vorbereiten
und Ausfallrisiken minimieren, mi-Fachverlag, Redline GmbH.
Sheffi, Y. (2007), The Resilient Enterprise: Overcoming Vulnerability for Competitive
Advantage, The MIT Press.
Shen, Z.-J. M.; Coullard, C. & Daskin, M. S. (2003), 'A joint location-inventory model',
Transportation Science 37(1), 40-55.
Sherbrooke, C. C. (1992), 'Multiechelon inventory systems with lateral supply', Naval
Research Logistics 39(1), 29-40.
Sheu, C. & Wacker, J. (1997), 'The effects of purchased parts commonality on
manufacturing lead time', International Journal of Operations and Production
Management 17(8), 725-745.
Shu, J.; Teo, C.-P. & Shen, Z.-J. M. (2005), 'Stochastic transportation-inventory network
design problem', Operations Research 53(1), 48-60.
Silver, E. A.; Pyke, D. F. & Peterson, R. (1998), Inventory Management and Production
Planning and Scheduling, John Wiley & Sons.
Simchi-Levi, D.; Kaminsky, P. & Simchi-Levi, E. (2008), Designing and Managing the
Supply Chain: Concepts, Strategies, and Case Studies, McGraw-Hill/Irwin.
Simpson, T. W.; Seepersad, C. C. & Mistree, F. (2001), 'Balancing commonality and
performance within the concurrent design of multiple products in a product family',
Concurrent Engineering 9(3), 177-190.
Slack, N. (1999), The Blackwell Encyclopedic Dictionary of Operations Management,
Wiley-Blackwell.
Slack, N.; Chambers, S. & Johnston, R. (2004), Operations Management, Prentice Hall.
Slikker, M.; Fransoo, J. & Wouters, M. (2001), 'Joint ordering in multiple newsvendor
situations: A game theoretical approach', BETA Working Paper 64, Eindhoven
University of Technology, 09/03/2001, http://alexandria.tue.nl/repository/books/553001
.pdf, accessed 09/23/2008.
Slikker, M.; Fransoo, J. & Wouters, M. (2005), 'Cooperation between multiple
newsvendors with transshipments', European Journal of Operational Research 167(2),
370-380.
Smith, S. A. & Agrawal, N. (2000), 'Management of multi-item retail inventory systems
with demand substitution', Operations Research 48(1), 50-64.
Snyder, K. (1995), 'New program set to provide ready inventory', Drug Topics 139(16), 19.
Snyder, L. V. & Shen, Z. J. M. (2006), 'Supply and demand uncertainty in multi-echelon
supply chains', technical report, P.C. Rossin College of Engineering and Applied
Sciences, Lehigh University, Bethlehem, PA, http://citeseerx.ist.psu.edu/viewdoc/down
load?doi=10.1.1.135.2642&rep=rep1&type=pdf, accessed 09/08/2010.
242
Bibliography
Snyder, L. V.; Daskin, M. S. & Teo, C.-P. (2007), 'The stochastic location model with risk
pooling', European Journal of Operational Research 179(3), 1221-1238.
Soanes, C. & Hawker, S. (2008), Compact Oxford English Dictionary of Current English,
Oxford University Press.
Sobel, M. J. (2008), Chapter 9: Risk pooling in Chhajed, D. & Lowe, T. J., eds., Building
Intuition: Insights from Basic Operations Management Models and Principles,
Söllner, A. (2008), Einführung in das Internationale Management, Gabler.
Sommerrock, F. (2009), Erfolgreiche Post-Merger-Integration bei öffentlichen
Institutionen: Fallstudienanalyse bei Sozialversicherungsträgern, Gabler.
Song, J.-S. & Zhao, Y. (2009), 'The value of component commonality in a dynamic
inventory system with lead times', Manufacturing & Service Operations Management
11(3), 493-508.
Sošić, G. (2006), 'Transshipment of inventories among retailers: Myopic vs. farsighted
stability', Management Science 52(10), 1493-1508.
Srinivasan, R.; Jayaraman, R.; Roundy, R. & Tayur, S. (1992), 'Procurement of common
components in a stochastic environment', Research Report RC-18580, IBM Thomas J.
Watson Research Center, Yorktown Heights, NY, 12/11/1992.
Staehle, W. (1994), Management: Eine verhaltenswissenschaftliche Perspektive, Vahlen.
Starbuck, W. H. (1981), A trip to view the elephants and rattlesnakes in the garden of
Aston in Van de Ven, A. H. & Joyce, W. F., eds., Perspectives on Organization Design
and Behavior, Wiley-Interscience, pp. 167-198.
Starbuck, W. H. (1982), 'Congealing oil: Inventing ideologies to justify acting ideologies
out', Joumal of Management Studies 19(1), 3-27.
Statistisches Bundesamt Deutschland (2009a), 'Beschäftigte, Umsatz, Aufwendungen,
Lagerbestände usw. im Handel - Fachserie 6 Reihe 4 - 1999', https://www-ec.de
statis.de/csp/shop/sfg/bpm.html.cms.cBroker.cls?CSPCHD=00300001000048zs1kmC0
00000HLZWPabwm_rSQ$IqeQ26IQ--&cmspath=struktur,AeltereTitel.csp&ID=10226
16, accessed 07/16/2009.
Statistisches Bundesamt Deutschland (2009b), 'Beschäftigte, Umsatz, Aufwendungen,
Lagerbestände usw. im Handel - Fachserie 6 Reihe 4 - 2000', vide supra.
Statistisches Bundesamt Deutschland (2009c), 'Beschäftigte, Umsatz, Aufwendungen,
Statistisches Bundesamt Deutschland (2009d), 'Beschäftigte, Umsatz, Aufwendungen,
Statistisches Bundesamt Deutschland (2009e), 'Beschäftigte, Umsatz, Aufwendungen,
Statistisches Bundesamt Deutschland (2009f), 'Beschäftigte, Umsatz, Aufwendungen,
Statistisches Bundesamt Deutschland (2009g), 'Beschäftigte, Umsatz, Aufwendungen,
243
Bibliography
Statistisches Bundesamt Deutschland (2009h), 'Beschäftigte, Umsatz, Aufwendungen,

Lagerbestände usw. im Handel - Fachserie 6 Reihe 4 - 2006', https://www-ec.destatis.de
/csp/shop/sfg/bpm.html.cms.cBroker.cls?cmspath=struktur,vollanzeige.csp&ID=102261
6, accessed 07/16/2009.
Statistisches Bundesamt Deutschland (2009i), Statistisches Jahrbuch 2009 für die
Bundesrepublik Deutschland, Statistisches Bundesamt Deutschland.
Statman, M. (1987), 'How many stocks make a diversified portfolio?', Journal of Financial
& Quantitative Analysis 22(3), 353-363.
Staudinger, M. (2007), Supply Management im industriellen Großanlagenbau, LIT Verlag.
Stein, A. (2000), Gestaltungsoptionen der Umschlagsknoten, in Klaus, P. & Krieger, W.,
eds., Gabler Lexikon Logistik: Management logistischer Netzwerke und Flüsse, Gabler,
pp. 485-488.
Stulman, A. (1987), 'Benefits of centralized stock for the multi-centre newsboy problem
with first come, first served allocation', Journal of the Operational Research Society
38(9), 827-832.
Su, X. (2008), 'Bounded rationality in newsvendor models', Manufacturing & Service
Operations Management 10(4), 566-589.
Sucky, E. (2009), 'The bullwhip effect in supply chains – An overestimated problem?',
Sugumaran, V. & Arogyaswamy, B. (2003), 'Measuring IT performance: "Contingency"
variables and value modes', Journal of Computer Information Systems 44(2), 79-86.
Sussams, J. (1986), 'Buffer stocks and the square root law', Focus on Physical Distribution
and Logistics Management 5(5), 8-10.
Swaminathan, J. M. & Lee, H. L. (2003), Chapter 5: Design for postponement in De Kok,
A. G. & Graves, S. C., eds., Handbooks in Operations Research and Management
Science: Supply Chain Management: Design, Coordination and Operation, Elsevier,
pp. 199-226.
Swaminathan, J. M. & Tayur, S. R. (1998), 'Managing broader product lines through
delayed differentiation using vanilla boxes', Management Science 44(12), S161-S172.
Swaminathan, J. M. (2001), 'Enabling customization using standardized operations',
California Management Review 43(3), 125-135.
Syntetos, A. A. & Boylan, J. E. (2001), 'On the bias of intermittent demand estimates',
International Journal of Production Economics 71(1-3), 457-466.
Syntetos, A. A. & Boylan, J. E. (2005), 'The accuracy of intermittent demand estimates',
International Journal of Forecasting 21(2), 303-314.
Syntetos, A. A. & Boylan, J. E. (2006), 'On the stock control performance of intermittent
demand estimators', International Journal of Production Economics 103(1), 36-47.
Syntetos, A. A.; Boylan, J. E. & Croston, J. D. (2005), 'On the categorisation of demand
patterns', Journal of the Operational Research Society 56(5), 495-503.
Taaffe, K.; Geunes, J. & Romeijn, H. E. (2008), 'Target market selection and marketing
effort under uncertainty: The selective newsvendor', European Journal of Operational
Research 189(3), 987-1003.
244
Bibliography
Tagaras, G. & Cohen, M. A. (1992), 'Pooling in two-location inventory systems with non-
negligible replenishment lead times', Management Science 38(8), 1067-1083.
Tagaras, G. & Vlachos, D. (2002), 'Effectiveness of stock transshipment under various
demand distributions and nonnegligible transshipment times', Production & Operations
Management 11(2), 183-198.
Tagaras, G. (1989), 'Effects of pooling on the optimization and service levels of two-
location inventory systems', IIE Transactions 21(3), 250-257.
Tagaras, G. (1999), 'Pooling in multi-location periodic inventory distribution systems',
Omega 27(1), 39-59.
Tallon, W. J. (1993), 'The impact of inventory centralization on aggregate safety stock:
The variable supply lead time case', Journal of Business Logistics 14(1), 185-203.
Taylor, D. A. (2004), Supply Chains: A Manager's Guide, Addison-Wesley.
Taylor, G. D., ed. (2008), Logistics Engineering Handbook, CRC Press Taylor & Francis
Group.
Tempelmeier, H. (2006), Inventory Management in Supply Networks, Books on Demand.
Teo, C. P.; Ou, J. & Goh, M. (2001), 'Impact on inventory costs with consolidation of
distribution centers', IIE Transactions 33(2), 99-110.
Teunter, R. & Sani, B. (2009), 'Calculating order-up-to levels for products with
intermittent demand', International Journal of Production Economics 118(1), 82-86.
Thayer, W. (2009), 'Kroger powers ahead', Refrigerated & Frozen Foods Retailer 7(3), 16-
26.
Thomas, D. & Hackman, S. (2003), 'A committed delivery strategy with fixed frequency
and quantity', European Journal of Operational Research 148(2), 363-373.
Thomas, D. J. & Tyworth, J. E. (2006), 'Pooling lead-time risk by order splitting: A critical
review', Transportation Research: Part E 42(4), 245-257.
Thomas, D. J. & Tyworth, J. E. (2007), 'Is pooling lead-time risk by splitting orders
simultaneously worthwhile?', Journal of Business Logistics 28(1), 169-193.
Thomas, L. C.; Possani, E. & Archibald, T. W. (2003), 'How useful is commonality?
Inventory and production decisions to maximize survival probability in start-ups', IMA
Journal of Management Mathematics 14(4), 305-320.
Thompson, B. (1994), 'The concept of statistical significance testing', Practical
Assessment, Research & Evaluation 4(5), retrieved 10/24/2009 from http://PAREonline.
net/getvn.asp?v=4\&n=5.
Thompson, B. (1996), 'AERA editorial policies regarding statistical significance testing:
Three suggested reforms', Educational Researcher 25(2), 26-30.
Thonemann, U. (2005), Operations Management: Konzepte, Methoden und Anwendungen,
Pearson Studium.
Thonemann, U. W. & Brandeau, M. L. (2000), 'Optimal commonality in component
design', Operations Research 48(1), 1-19.
Tiacci, L. & Saetta, S. (2009), 'An approach to evaluate the impact of interaction between
demand forecasting method and stock control policy on the inventory system
performances', International Journal of Production Economics 118(1), 63-71.
245
Bibliography
Tomlin, B. & Wang, Y. (2005), 'On the value of mix flexibility and dual sourcing in
unreliable newsvendor networks', Manufacturing & Service Operations Management
7(1), 37-57.
Treece, J. B. (1996), 'Saturn picks Japan helper', Automotive News 71(5675), 43.
Tsubone, H.; Matsuura, H. & Satoh, S. (1994), 'Component part commonality and process
flexibility effects on manufacturing performance', International Journal of Production
Research 32(10), 2479-2493.
Turner, N. (2001), 'Choosing the most appropriate warehouse management system',
Logistics & Transport Focus 3(7), 30-33.
Tyagi, R. & Das, C. (1998), 'Extension of the square-root law for safety stock to demands
with unequal variances', Journal of Business Logistics 19(2), 197-203.
Tyagi, R. & Das, C. (1999), 'Grouping customers for better allocation of resources to serve
correlated demands', Computers and Operations Research 26(10-11), 1041-1058.
Tyworth, J. E. & Ruiz-Torres, A. (2000), 'Transportation's role in the sole versus dual-
sourcing decisions', International Journal of Physical Distribution and Logistics
Management 30(2), 128-136.
U.S. Census Bureau (2009), 'Latest monthly wholesale trade report', http://www2.census.
gov/wholesale/xls/mwts/historic1.xls, accessed 07/16/2009.
Ulrich, K. & Ellison, D. J. (1999), 'Holistic customer requirements and the design-select
decision', Management Science 45(5), 641-658.
Upton, D. M. (1994), 'The management of manufacturing flexibility', California
Management Review 36(2), 72-89.
Upton, D. M. (1995), 'What really makes factories flexible?', Harvard Business Review
73(4), 74-84.
Vahrenkamp, R. (2000), Logistikmanagement, Oldenbourg.
Vakharia, A.; Parmenter, D. A. & Sanchez, S. (1996), 'The operating impact of parts
commonality', Journal of Operations Management 14(1), 3-18.
Van Hoek, R. I. & Van Dierdonck, R. (2000), 'Postponed manufacturing supplementary to
transportation services?', Transportation Research: Part E 36(3), 205-217.
Van Hoek, R. I. (1997), 'Postponed manufacturing: A case study in the food supply chain',
Supply Chain Management: An International Journal 2(2), 63-75.
Van Hoek, R. I. (1998a), 'Reconfiguring the supply chain to implement postponed
manufacturing', International Journal of Logistics Management 9(1), 95-110.
Van Hoek, R. I. (1998b), 'Logistics and virtual integration: Postponement, outsourcing and
the flow of information', International Journal of Physical Distribution & Logistics
Management 28(7), 508-523.
Van Hoek, R. I. (2000a), 'The thesis of leagility revisited', International Journal of Agile
Management Systems 2(3), 196-201.
Van Hoek, R. I. (2000b), 'The role of third-party logistics providers in mass customization',
Van Hoek, R. I. (2001), 'The rediscovery of postponement: A literature review and
directions for research', Journal of Operations Management 19(2), 161-184.
246
Bibliography
Van Hoek, R. I.; Commandeur, H. R. & Vos, B. (1998), 'Reconfiguring logistics systems
through postponement strategies', Journal of Business Logistics 19(1), 33-54.
Van Hoek, R. I.; Peelen, E. & Commandeur, H. R. (1999a), 'Achieving mass customization
through postponement: A study of international changes', Journal of Market Focused
Management 3(3-4), 353-368.
Van Hoek, R. I.; Vos, B. & Commandeur, H. R. (1999b), 'Restructuring European supply
chains by implementing postponement strategies', Long Range Planning 32(5), 505-
518.
Van Mieghem, J. A. & Dada, M. (1999), 'Price versus production postponement: Capacity
and competition', Management Science 45(12), 1631-1649.
Van Mieghem, J. A. (2004), 'Note–Commonality strategies: Value drivers and equivalence
with flexible capacity and inventory substitution', Management Science 50(3), 419-424.
Van Mieghem, J. A. (2007), 'Risk mitigation in newsvendor networks: Resource
diversification, flexibility, sharing, and hedging', Management Science 53(8), 1269-
1288.
Van Mieghem, J. A. (2008), Operations Strategy: Principles and Practice, Dynamic Ideas.
Verkoeijen, P. C. F. M. & deHaas, R. C. T. (1998), 'Virtual warehousing, an inventory
(virtual warehousing, een inventarisatie)', technical report, NASA no. 19990036800,
NATO Maintenance and Supply Agency (NASA).
Vidyarthi, N.; Çelebi, E.; Elhedhli, S. & Jewkes, E. (2007), 'Integrated production-
inventory-distribution system design with risk pooling: Model formulation and heuristic
solution', Transportation Science 41(3), 392-408.
Vockell, E. L. & Asher, J. W. (1995), Educational Research, Allyn & Bacon.
Voorhees, R. D. & Sharp, M. K. (1978), 'The principles of logistics revisited',
Transportation Journal 18(1), 69-84.
Wacker, J. G. & Treleven, M. (1986), 'Component part standardization: An analysis of
commonality sources and indices', Journal of Operations Management 6(2), 219-244.
Wacker, J. G. (2004), 'A theory of formal conceptual definitions: Developing theory-
building measurement instruments', Journal of Operations Management 22(6), 629-650.
Wagner, G. R. (1997), Betriebswirtschaftliche Umweltökonomie, Lucius & Lucius.
Wagner, S. M. & Bode, C. (2008), 'An empirical examination of supply chain performance
along several dimensions of risk', Journal of Business Logistics 29(1), 307-325.
Walleigh, R. (1989), 'Product design for low-cost manufacturing', Journal of Business
Strategy 10(4), 37-41.
Waller, M. A.; Dabholkar, P. A. & Gentry, J. J. (2000), 'Postponement, product
customization, and market-oriented supply chain management', Journal of Business
Logistics 21(2), 133-160.
Walonick, D. (2009), Survival Statistics, StatPac, Inc.
Wang, H. W.; Chen, R. Q. & Li, Y. (2006), 'Case study on the application of postponement
strategy and managerial insights', Asia Pacific Management Review 11(3), 141-153.
Wang, Y.; Jiang, L. & Shen, Z. J. (2004), 'Channel performance under consignment
contract with revenue sharing', Management Science 50(1), 34-47.
247
Bibliography
Wanke, P. (1999), 'Formalizando uma Política de Estoques para a Cadeia de Suprimentos',

Revista Tecnologística 5(48), 22-29.
Wanke, P. F. & Saliby, E. (2009), 'Consolidation effects: Whether and how inventories
should be pooled', Transportation Research: Part E 45(5), 678-692.
Wanke, P. F. & Zinn, W. (2004), 'Strategic logistics decision making', International
Journal of Physical Distribution & Logistics Management 34(6), 466-478.
Wanke, P. F. (2009), 'Consolidation effects and inventory portfolios', Transportation
Research: Part E 45(1), 107-124.
Wannenwetsch, H. (2009), Integrierte Materialwirtschaft und Logistik: Beschaffung,
Logistik, Materialwirtschaft und Produktion, Springer.
Watson, M. & Morton, J. (2000), 'Designing perfect distribution channels: Software helps
determine optimal plan', Parcel Shipping & Distribution Magazine, April, http://e-
opt.informs.org/resources/uploads/ACF2CBF.pdf, accessed 03/28/2008.
Wazed, M. A.; Ahmed, S. & Yusoff, N. (2008), 'Commonality models in manufacturing
resources planning: State-of-the-art and future directions', European Journal of
Scientific Research 23(3), 421-435.
Wazed, M. A.; Ahmed, S. & Yusoff, N. (2009), 'Commonality and its measurement in
manufacturing resources planning', Journal of Applied Sciences 9(1), 69-78.
Weber, M. (1980), Wirtschaft und Gesellschaft: Grundriß der verstehenden Soziologie,
Mohr.
Wee, K. E. & Dada, M. (2005), 'Optimal policies for transshipping inventory in a retail
network', Management Science 51(10), 1519-1533.
Wegner, U. (1996), Einführung in das Logistikmanagement: Prozesse, Strukturen,
Anwendungen, Gabler.
Weisstein, E. W. (2010), 'Variance', http://mathworld.wolfram.com/Variance.html, access
02/26/2010.
Weng, Z. K. & McClurg, T. (2003), 'Coordinated ordering decisions for short life cycle
products with uncertainty in delivery time and demand', European Journal of
Operational Research 151(1), 12.
Weng, Z. K. (1998), 'Managing production with flexible capacity deployment for serial
multi-stage manufacturing systems', European Journal of Operational Research 109(3),
587-598.
Weng, Z. K. (1999), 'Risk-pooling over demand uncertainty in the presence of product
modularity', International Journal of Production Economics 62(1), 75-85.
Wessiepe, K. (2009), 'Das Postleitzahlen-Diagramm 3.7', http://www.klaus-wessiepe.de/
plz.htm, accessed 07/18/2009.
White, S. C. (2005), 'Freight consolidation', Special Report, Council of Supply Chain
Management Professionals, April 2005, cscmp.org/downloads/public/resources/freight
con.pdf, accessed 09/14/2009.
Whybark, D. C. (1989), Case 4-3: International plow, in 'International Operations
Management: A Selection of IMEDE Cases', Irwin, pp. 267-274.
248
Bibliography
Wildemann, H. (2008), Produktkannibalisierung erkennen und vermeiden in Specht, D.,

ed., Produkt- und Prozessinnovationen in Wertschöpfungsketten: Tagungsband der
Herbsttagung 2007 der Wissenschaftlichen Kommission Produktionswirtschaft im VHB,
Gabler, pp. 71-83.
Willemain, T. R.; Smart, C. N.; Shockor, J. H. & DeSautels, P. A. (1994), 'Forecasting
intermittent demand in manufacturing: A comparative evaluation of Croston's method',
International Journal of Forecasting 10(4), 529-538.
Williams, J. (1975), 'Food distribution costs: Results of an inter-firm study of wholesale
transportation and warehousing costs', technical report, National Materials Handling
Center Cranfield, Cranfield, UK.
Wisner, J. D.; Tan, K.-C. & Leong, G. K. (2009), Principles of Supply Chain Management:
A Balanced Approach, South-Western Cengage Learning.
Wong, C. Y. & Hvolby, H.-H. (2007), 'Coordinated responsiveness for volatile toy supply
chains', Production Planning & Control 18(5), 407-419.
Wong, H. (2005), 'Pooling of repairable spare parts: A study on inventory policies', 4OR: A
Quarterly Journal of Operations Research 3(3), 253-256.
Wong, H.; Cattrysse, D. & van Oudheusden, D. (2005a), 'Stocking decisions for repairable
spare parts pooling in a multi-hub system', International Journal of Production
Economics 93-94(1), 309-317.
Wong, H.; van Houtum, G. J.; Cattrysse, D. & van Oudheusden, D. (2005b), 'Simple,
efficient heuristics for multi-item multi-location spare parts systems with lateral
transshipments and waiting time constraints', Journal of the Operational Research
Society 56(12), 1419-1430.
Wong, H.; van Oudheusden, D. & Cattrysse, D. (2007a), 'Cost allocation in spare parts
inventory pooling', Transportation Research: Part E 43(4), 370-386.
Wong, H.; van Oudheusden, D. & Cattrysse, D. (2007b), 'Two-echelon multi-item spare
parts systems with emergency supply flexibility and waiting time constraints', IIE
Transactions 39(11), 1045-1057.
Wong, H.; Wikner, J. & Naim, M. (2009), 'Analysis of form postponement based on
optimal positioning of the differentiation point and stocking decisions', International
Journal of Production Research 47(5), 1201-1224.
Woolsey, G. (1988), 'A requiem for the EOQ: An editorial', Production and Inventory
Management Journal 29(3), 68-72.
Wright, P. B. (1994), 'The humana case and other tax issues', Risk Management
(00355593) 41(5), 77-78.
Xu, K. & Evers, P. T. (2003), 'Managing single echelon inventories through demand
aggregation and the feasibility of a correlation matrix', Computers & Operations
Research 30(2), 297-309.
Xu, K.; Evers, P. T. & Fu, M. C. (2003), 'Estimating customer service in a two-location
continuous review inventory model with emergency transshipments', European Journal
of Operational Research 145(3), 569-585.
Yanagi, S. & Sasaki, M. (1992), 'An approximation method for the problem of a
repairable-item inventory system with lateral resupply', IMA Journal of Mathematics
Applied in Business and Industry 3(4), 305-314.
249
Bibliography
Yang, B.; Burns, N. & Backhouse, C. (2004), 'Management of uncertainty through

postponement', International Journal of Production Research 42(6), 1049-1064.
Yang, B.; Burns, N. D. & Backhouse, C. J. (2005a), 'The application of postponement in
industry', IEEE Transactions on Engineering Management 52(2), 238-248.
Yang, B.; Burns, N. D. & Backhouse, C. J. (2005b), 'An empirical investigation into the
barriers to postponement', International Journal of Production Research 43(5), 991-
1005.
Yang, B.; Yang, Y. & Wijngaard, J. (2007), 'Postponement: An inter-organizational
perspective', International Journal of Production Research 45(4), 971-988.
Yang, H. & Schrage, L. (2003), 'An inventory anomaly: Risk pooling may increase
inventory', technical report, Graduate School of Business, University of Chicago, IL.
Yang, H. & Schrage, L. (2009), 'Conditions that cause risk pooling to increase inventory',
Yang, J. & Qin, Z. (2007), 'Capacitated production control with virtual lateral
transshipments', Operations Research 55(6), 1104-1119.
Yeung, J. H. Y.; Selen, W.; Deming, Z. & Min, Z. (2007), 'Postponement strategy from a
supply chain perspective: Cases from China', International Journal of Physical
Yu, J. & Cooper, H. (1983), 'A quantitative review of research design effects on response
rates to questionnaires', Journal of Marketing Research 20(1), 36-44.
Yu, Y.; Benjaafar, S. & Gerchak, Y. (2008), 'Capacity pooling and cost sharing among
independent firms in the presence of congestion', 02/27/2008, http://www.me.umn.edu/
research/faculty/pdf/ybg-2-27-08.pdf, accessed 03/28/2008.
Zhang, D. & Cooper, W. L. (2005), 'Revenue management for parallel flights with
customer-choice behavior', Operations Research 53(3), 415-431.
Zhang, J. (2005), 'Transshipment and its impact on supply chain members' performance',
Zhang, J. (2009), 'Cost allocation for joint replenishment models', Operations Research
57(1), 146-156.
Zhang, Q.; Vonderembse, M. A. & Lim, J.-S. (2002), 'Value chain flexibility: A dichotomy
of competence and capability', International Journal of Production Research 40(3),
561-583.
Zhao, H.; Deshpande, V. & Ryan, J. K. (2006), 'Emergency transshipment in decentralized
dealer networks: When to send and accept transshipment requests', Naval Research
Logistics 53(6), 547-567.
Zhao, H.; Deshpande, V. & Ryan, J. K. (2005), 'Inventory sharing and rationing in
decentralized dealer networks', Management Science 51(4), 531-547.
Zhao, H.; Ryan, J. K. & Deshpande, V. (2008), 'Optimal dynamic production and
inventory transshipment policies for a two-location make-to-stock system', Operations
Research 56(2), 400-410.
Zhao, X. & Atkins, D. (2009), 'Transshipment between competing retailers', IIE
250
Bibliography
Zhou, L. & Grubbström, R. W. (2004), 'Analysis of the effect of commonality in multi-

level inventory systems applying MRP theory', International Journal of Production
Economics 90(2), 251-263.
Zhou, L. G. & Lau, H. S. (1992), 'Reducing inventory costs and choosing suppliers with
order splitting', Journal of the Operational Research Society 43(10), 1003-1008.
Zinn, W. & Bowersox, D. J. (1988), 'Planning physical distribution with the principle of
postponement', Journal of Business Logistics 9(2), 117-136.
Zinn, W. (1990), 'Developing heuristics to estimate the impact of postponement on safety
stock', International Journal of Logistics Management 1(2), 11-16.
Zinn, W.; Levy, M. & Bowersox, D. J. (1989), 'Measuring the effect of inventory
centralization/decentralization on aggregate safety stock: The "square root law"
revisited', Journal of Business Logistics 10(1), 1-14.
Zinn, W.; Levy, M. & Bowersox, D. J. (1990), 'On assumed assumptions and the inventory
centralization/decentralization issue', Journal of Business Logistics 11(2), 139-142.
251

Dissertation Gerald Oeser

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dissertation Gerald Oeser

Uploaded by

Copyright:

Available Formats

Methods of Risk Pooling in Business Logistics

and Their Application

zur Erlangung des akademischen Grades

John Allen Paulos,

Prof. Dr. Dr. h.c. Knut Richter

List of Figures .................................................................................................................... X

List of Tables ..................................................................................................................... XI

List of Abbreviations, Signs, and Symbols ................................................................... XII

Abstract ................................................................................................................... XIV

1.1 Problem: Growing Uncertainty in Business Logistics and

2 Risk Pooling in Business Logistics ............................................................. 6

2.1 Business Logistics Risk Pooling Research ................................................... 6

3 Methods of Risk Pooling ........................................................................... 24

3.1 Storage: Inventory Pooling ......................................................................... 27

3.5 Sales and Distribution ................................................................................. 51

4 Choosing Suitable Risk Pooling Methods ............................................... 53

4.1 Contingency Theory .................................................................................... 53

5 Applying Risk Pooling at Papierco .......................................................... 74

5.1 Papierco ....................................................................................................... 74

6 Conclusion ............................................................................................... 126

Appendix A: A Survey on Risk Pooling Knowledge and Application in

1 Introduction ............................................................................................. 135

1.1 Motivation: Scarce Survey Research on Risk Pooling ............................. 135

2 The Survey ............................................................................................... 142

2.1 Research Design ........................................................................................ 142

3 Critical Appraisal .................................................................................... 166

4 Questionnaire .......................................................................................... 169

5 Answers .................................................................................................... 170

Appendix C: What Causes the Savings in Regular Stock through Centralization

Appendix D: Tables ....................................................................................................... 178

Bibliography .................................................................................................................... 204

Figure 2.1: Number of Publications on Risk Pooling in Business Logistics .................. 7

Table 3.1: Risk Pooling Methods' Building Blocks ..................................................... 26

PCE portfolio cost effect

1.1 Problem: Growing Uncertainty in Business Logistics and

1.3 Object and Method

The research object (German: Erfahrungsgegenstand) is an empirical phenomenon (a part of

1.4 Structure and Contents

2.1 Business Logistics Risk Pooling Research

Order Splitting (37)

Figure 2.1: Number of Publications on Risk Pooling in Business Logistics

2.2 Placing Risk Pooling in the Supply Chain,

Lead Time Uncertainty

Material, product, and information flow

Risk pooling methods

Figure 2.2: Placing Risk Pooling in Business Logistics

Figure 2.3: Important Value Activities Using Risk Pooling Methods

2.3 Defining Risk Pooling

We define risk pooling in business logistics as consolidating individual variabilities

2.4 Explaining Risk Pooling

(2) Random variables xi cannot mutually balance their fluctuations, if at least

2.5 Characteristics of Risk Pooling

 retailers in pooling over the outside-supplier lead time or centralized ordering135

After a thorough literature review we conclude that apart from

(7) postponement, delayed (product) differentiation,202 process standardiza-

Risk Pooling Methods

Table 3.1: Risk Pooling Methods' Building Blocks

3.1 Storage: Inventory Pooling

Inventory centralization can reduce expected costs in a cost minimization model246 or

Centralization can lead to economies of massed reserves, if it is used to lower system-

Researchers questioning Maister's (1976) SRL challenge its assumptions illogically297, do

3.2.1 Virtual Pooling

Emergency or reactive327 lateral transshipments (ELT) transfer products from a loca-

3.3.1 Centralized Ordering

, 018 *!18

% &. &,

/.9 &)#