Evaluation of Cooperative Planning in Supply Chains: An Empirical Approach of the European Automotive Industry (Produktion und Logistik) 9783835004313, 383500431X

Citation preview

---

Luis Martín Díaz Evaluation of Cooperative Planning in Supply Chains

GABLER EDITION WISSENSCHAFT Produktion und Logistik Herausgegeben von Professor Dr. Wolfgang Domschke, Technische Universität Darmstadt, Professor Dr. Andreas Drexl, Universität Kiel, Professor Dr. Bernhard Fleischmann, Universität Augsburg, Professor Dr. Hans-Otto Günther, Technische Universität Berlin, Professor Dr. Christoph Haehling von Lanzenauer, Freie Universität Berlin, Professor Dr. Karl Inderfurth, Universität Magdeburg, Professor Dr. Klaus Neumann, Universität Karlsruhe, Professor Dr. Christoph Schneeweiß, Universität Mannheim (em.), Professor Dr. Hartmut Stadtler, Technische Universität Darmstadt, Professor Dr. Horst Tempelmeier, Universität zu Köln, Professor Dr. Gerhard Wäscher, Universität Magdeburg

Kontakt: Professor Dr. Hans-Otto Günther, Technische Universität Berlin, FG BWL – Produktionsmanagement, Wilmersdorfer Str. 148, 10585 Berlin

Diese Reihe dient der Veröffentlichung neuer Forschungsergebnisse auf den Gebieten der Produktion und Logistik. Aufgenommen werden vor allem herausragende quantitativ orientierte Dissertationen und Habilitationsschriften. Die Publikationen vermitteln innovative Beiträge zur Lösung praktischer Anwendungsprobleme der Produktion und Logistik unter Einsatz quantitativer Methoden und moderner Informationstechnologie.

Luis Martín Díaz

Evaluation of Cooperative Planning in Supply Chains An Empirical Approach of the European Automotive Industry

With a foreword by Prof. Dr. Peter Buxmann

Deutscher Universitäts-Verlag

Bibliografische Information Der Deutschen Bibliothek Die Deutsche Bibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über abrufbar.

Dissertation Technische Universität Darmstadt, 2005 D17

1. Auflage Mai 2006 Alle Rechte vorbehalten © Deutscher Universitäts-Verlag | GWV Fachverlage GmbH, Wiesbaden 2006 Lektorat: Brigitte Siegel / Nicole Schweitzer Der Deutsche Universitäts-Verlag ist ein Unternehmen von Springer Science+Business Media. www.duv.de Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlags unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Die Wiedergabe von Gebrauchsnamen, Handelsnamen, Warenbezeichnungen usw. in diesem Werk berechtigt auch ohne besondere Kennzeichnung nicht zu der Annahme, dass solche Namen im Sinne der Warenzeichen- und Markenschutz-Gesetzgebung als frei zu betrachten wären und daher von jedermann benutzt werden dürften. Umschlaggestaltung: Regine Zimmer, Dipl.-Designerin, Frankfurt/Main Druck und Buchbinder: Rosch-Buch, Scheßlitz Gedruckt auf säurefreiem und chlorfrei gebleichtem Papier Printed in Germany ISBN-10 3-8350-0431-X ISBN-13 978-3-8350-0431-3

To my wife Silke, for her all-embracing support, and to my children Adrián and Fabio, for requiring only so much time that I could still finish this work.

Foreword

The acknowledgement that a network of cooperating companies, e.g. a supply chain, could be more successful in achieving competitive advantage than individual businesses, constitutes one of the most significant paradigm shifts in modern business management as it leaves behind the notion of adversarial companies engaged in fierce competition with one another in order to gain a competitive advantage. The “Survival of the Fittest” in what has been called “The Era of Network Competition” depends on how well companies are able to structure, coordinate, and manage relationships with their business partners. In their search for the best possible place under the sun, companies have redefined their understanding of cooperation and have not only improved the efficiency of cooperation with partners (e.g. suppliers, customers, and complementors) but also discovered the existence of synergies with competitors. The apparently paradoxical situation of cooperating with competitors (or it might be thought of as competing with cooperating partners) shows that collaboration is a widespread approach at all levels of strategic management. In light of these thoughts, the question arises why there are still companies that neither cooperate with business partners nor with competitors; although it seems obvious that this could be advantageous. This work attempts to offer an answer to this question. In so doing, it explores the reasons why companies do not engage in cooperation despite the fact that it would appear to be advantageous for them to do so. A very extensive empirical study within the European automotive industry identifies trust and issues of complexity among the significant causes for not cooperating.

However, the most important motive for the

unwillingness to cooperate is the evaluation of the actual benefits implied in the cooperation. Uncooperative companies seem to be asking themselves if the coordination and integration efforts incurred within cooperation actually correspond to adequate benefits.

VIII

Foreword

In order to provide an answer, Luis Martín Díaz analyzes economic approaches in evaluating alternative forms of cooperation in supply chains. Firstly, the author shows in a case study of Audi AG how existing cooperative environments, such as a supply chain of the automotive industry, can be evaluated using the bullwhip effect as an indicator. The thorough and detailed analysis reveals a considerable bullwhip effect caused by increased transportation times and unequal supply call-off strategies at different stages of the supply chain. Secondly, Luis Martín Díaz presents an innovative software prototype that assesses the benefits of different cooperation forms at the planning level. Towards that end, the prototype evaluates the impact of alternative decision structures in the results of transportation planning and the bullwhip effect.

The

inventive combination of different decision structures with specific logistics planning methods permits a precise estimate of the potential benefits derived from a collaboration at an early stage of the decision process. The work of Luis Martín Díaz contains many good and innovative ideas and represents an outstanding contribution to research. I hope that it attains widespread acceptance and that the reader will enjoy it as much as I did.

Professor Dr. Peter Buxmann

Acknowledgements

This dissertation presents economic approaches, quantitative methods, and a software prototype in order to quantify the benefits of cooperation in supply chains. The limits of my analysis become apparent when attempting to quantify the benefits of the help of the people who contributed to the successful completion of this work. This invaluable collaboration can only be addressed in general terms and with great gratitude; thus, making it virtually impossible to put into words. However, I would like to attempt it. First of all, I would really like to express my deepest gratitude to Professor Dr. Peter Buxmann for being more than the best dissertation supervisor a doctoral candidate could imagine having. He offered me all of his knowledge and experience, made my goals his, and gave me both orientation and freedom in the right dose, providing me with all the resources I needed. I also would like to thank the German Research Foundation (DFG) for supporting my research in the context of the SKILNET research project with the grant BU 1098/1-1/2. Professor Dr. Wolfgang Domschke deserves my most sincere thanks for his review of my work as well as Professor Dr. Dr. Oskar Betsch, Professor Dr. Volker Caspari, and Professor Dr. Jochen Marly for their participation in the disputation. I am indebted to Professor Warren B. Powell for making it possible to do part of my research at Princeton University. Many thanks also to the staff members at Castle Laboratory, especially to Hugo Passos Simão for the fantastic collaboration during my stay at Princeton. Dr. Kai Salzmann deserves many thanks for giving me an insight into the supply chain of Audi AG. Many thanks also to Professor Dr. Axel Kimms for the many hours of prolific discussions on the topic of quantitative analysis and for his excellent suggestions.

X

Acknowledgements

I want to also express my great appreciation to Dr. Erik Wüstner for being a fantastic colleague all these years and for contributing with thoroughness and inestimable hints to the success of my work. Special thanks go to Kristina Wolf and Anette von Ahsen for their support within the empirical study, and to Tina Werthmann for her unfailing contribution to the case study. Antje Heidl, Jens Kühner, Michael Meyer, and Andreas Knabe have earned my gratitude for their valuable support in the development process of the software prototype. Many kind thanks also go to Martina Lohmann-Hinner for carefully reading the manuscript and for her professionalism and generous effort. Thanks also to Nicole Schweitzer at Gabler for the smooth and friendly publication process. My deepest appreciation goes to my wife Silke for her continuous and affectionate help and for always being the splendid support I needed. I am also very grateful to my parents-in-law Clara and Hartmut Strate for providing a helpful hand at all times and to my parents for their excellent aid during all stages of my education.

Luis Martín Díaz

Table of Contents

List of Figures.....................................................................................................................................................XV List of Tables .................................................................................................................................................... XXI Abbreviations ................................................................................................................................................ XXVII 1

2

Introduction....................................................................................................................................................1 1.1

Motivation and Research Questions ............................................................................................................4

1.2

Structure of the Dissertation .......................................................................................................................6

Inter-Organizational Cooperation and Supply Chain Management ..........................................................9 2.1

Inter-Organizational Cooperation..............................................................................................................10 2.1.1

2.2

2.3

Definition of Cooperation ................................................................................................................11

2.1.2

Definition of Inter-Organizational Cooperation................................................................................14

2.1.3

Forms of Inter-Organizational Cooperation......................................................................................16

2.1.3.1

Redistributive and Reciprocal Inter-Organizational Cooperation...............................................................17

2.1.3.2

X and Y Inter-Organizational Cooperation...............................................................................................19

2.1.3.3

Rotering Matrix for Identifying the Type of Inter-Organizational Cooperation .........................................21

Supply Chain Management........................................................................................................................22 2.2.1

Supply Chain Management as a Field of Research and of Practical Endeavors ..................................22

2.2.2

Defining Supply Chain Management.................................................................................................24

2.2.3

The Objectives of Supply Chain Management...................................................................................28

2.2.4

Issues Related to Cooperation in the Context of Supply Chain Management.....................................32

2.2.4.1

Co-opetition – A Concept Describing Simultaneous Cooperation and Competition..................................33

2.2.4.2

The Bullwhip Effect – A Frequent Problem in Supply Chains with Lower Degree of Cooperation ...........35

2.2.4.2.1

Demand Forecast Updating..............................................................................................................36

2.2.4.2.2

Order Batching ................................................................................................................................37

2.2.4.2.3

Price Fluctuation ..............................................................................................................................38

2.2.4.2.4

Rationing and Shortage Gaming .......................................................................................................38

Logistics Planning as Object of Inter-Organizational Cooperation.............................................................39 2.3.1

Business Logistics – a Supply Chain Management Process ................................................................39

2.3.2

Logistics Planning as a Hierarchical Planning Problem......................................................................42

XII

3

Table of Contents 2.3.3

Inter-Organizational Logistics Planning in Supply Chains as a Hierarchical Planning Problem ..........46

2.3.4

Inter-Organizational Planning – The Approach of Wyner and Malone..............................................47

Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry .............51 3.1

3.2

Cooperative Transportation in Supply Chains............................................................................................51 3.1.1

An Exemplary Decision Category in Logistics: Transportation..........................................................51

3.1.2

Selected Cooperative Scenarios for Transportation ...........................................................................55

3.1.2.1

Cooperation Scenario I: Engaging in a Logistics Alliance ..........................................................................56

3.1.2.2

Cooperation scenario II: Supply Chain-Wide Container Management .......................................................60

3.1.2.3

Cooperation Scenario III: Selling Excess Transportation Capacity to Other Companies............................60

3.1.2.4

Cooperation Scenario IV: Joint Ownership of Transportation Capacity ....................................................62

3.1.2.5

Cooperation Scenario V: Multi-Stop Shipping and Sequenced Loading .....................................................62

3.1.2.6

Cooperation Scenario VI: Merge-in-Transit and Sequenced Loading.........................................................64

3.1.2.7

Cooperation Scenario VII: Cross Docking and Sequenced Loading ..........................................................67

3.1.2.8

Cooperation Scenario VIII: Sequenced Delivery .......................................................................................69

SCM Software as an Instrument for Cooperative Planning in Supply Chains – An Explorative Survey on the European Automotive Industry...........................................................................................71 3.2.1

Information Sharing as Premise for Cooperation in Supply Chains ...................................................71

3.2.2

Goals of the Survey ..........................................................................................................................78

3.2.3

Research Design ...............................................................................................................................79

3.2.4

Cooperation in the European Automotive Industry ..........................................................................80

3.2.4.1

Fields of Cooperation...............................................................................................................................81

3.2.4.2

Collaborative Planning .............................................................................................................................81

3.2.5

The Status Quo of Using Supply Chain Management Software .................................................................84

3.2.5.2

Goals of Implementing Supply Chain Management Software....................................................................86

3.2.5.3

Evaluation of Benefits from Using Supply Chain Management Software...................................................88

3.2.5.4

Supply Chain Management Software and Network Effects in the European Automotive Industry ............90

3.2.6 4

Supply Chain Management Software in the European Automotive Industry......................................84

3.2.5.1

Summary of Results ..........................................................................................................................94

The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG ............................97 4.1

Exposés of Companies in the Supply Chain of the V8 4.0l Diesel Engine..................................................98 4.1.1

4.2

Audi AG...........................................................................................................................................98

4.1.1.1

General Overview of Audi AG.................................................................................................................98

4.1.1.2

Supply Chain Related Overview of Audi AG ..........................................................................................101

4.1.2

Audi Hungaria Motor Kft. ..............................................................................................................104

4.1.3

TCG Unitech Systemtechnik ..........................................................................................................104

4.1.4

Gustav Wahler GmbH u. Co. KG ..................................................................................................105

Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine ..............................105 4.2.1

Description of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine ..........................................106

4.2.2

Analysis of Inventory Levels in the Supply Chain of the Audi A8 V8 4.0l Diesel Engine.................114

4.2.3

Analysis of Orders Placed by the Companies in the Supply Chain of the Audi A8 V8 4.0l Diesel Engine .................................................................................................................................121

4.3

Evaluation of the Audi AG Supply Chain................................................................................................129

Table of Contents

XIII

4.3.1

The Bullwhip Effect in the Supply Chain of the Audi A8 V8 4.0l Diesel Engine .............................129

4.3.2

Evaluation of the Cooperation Scenarios for Transportation in the Supply Chain of the Audi A8 V8 4.0l Diesel Engine................................................................................................................135

4.3.2.1

Cooperation Scenario I: Engaging in a Logistics Alliance ........................................................................135

4.3.2.2

Cooperation Scenario II: Supply Chain-wide Container Management......................................................138

4.3.2.3

Cooperation Scenario III: Selling Excess Transportation Capacity to Other Companies..........................139

4.3.2.4

Cooperation Scenario IV: Joint Ownership of Transportation Capacity ..................................................140

4.3.2.5

Cooperation Scenario V: Multi-Stop Shipping and Sequenced Loading ...................................................140

4.3.2.6

Cooperation Scenario VI: Merge-in-Transit and Sequenced Loading.......................................................141

4.3.2.7

Cooperation Scenario VII: Cross Docking and Sequenced Loading ........................................................141

4.3.2.8

Cooperation scenario VIII: Sequenced delivery.......................................................................................142

4.3.3

Evaluation of the Implementation of Supply Chain Monitoring in the Audi AG Supply Chain: Real-time Exchange of Information on Capacity, Inventory, and Demand......................................143

4.4 5

Summary of Results ................................................................................................................................152

SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains ......159 5.1

Prototypical Implementation ...................................................................................................................161 5.1.1

The SCOptimizer Architecture .......................................................................................................161

5.1.2

Prototypical Implementation of the Evaluation of Cooperative Distribution Planning with the SCOptimizer...................................................................................................................................169

5.1.2.1

Planning Background .............................................................................................................................169

5.1.2.2

Description from the Planner’s Point of View ........................................................................................174

5.1.3

5.2

Prototypical Implementation of the Evaluation of the Bullwhip Effect ...........................................190

5.1.3.1

Planning Background .............................................................................................................................190

5.1.3.2

Description from the Planner’s Point of View ........................................................................................198

Computational Study on Cooperative Distribution – An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer.............................................................................................................222 5.2.1

Approach of the Computational Study............................................................................................223

5.2.2

Selected Results of the Computational Study...................................................................................228

5.2.2.1

Total Costs.............................................................................................................................................229

5.2.2.2

Vehicle Costs .........................................................................................................................................236

5.2.2.3

Distance Costs .......................................................................................................................................238

5.2.2.4

Capacities...............................................................................................................................................240

5.2.2.5

Relative Results ......................................................................................................................................244

5.2.3 6

Summary of Results ........................................................................................................................248

Summary and Conclusions........................................................................................................................253 6.1

Summary of the Findings and Implications..............................................................................................254

6.2

Outlook and Further Research ................................................................................................................257

References...........................................................................................................................................................261

List of Figures

Figure 1:

Delimitation of the Coordination Forms Market-Cooperation-Hierarchy ..................12

Figure 2:

Redistributive Cooperation .................................................................................................18

Figure 3:

Reciprocal Cooperation .......................................................................................................19

Figure 4:

Y Cooperation.......................................................................................................................20

Figure 5:

X Cooperation.......................................................................................................................21

Figure 6:

Rotering Matrix for Classifying Inter-Organizational Cooperation ..............................21

Figure 7:

Levels of Supply Chain Complexity...................................................................................27

Figure 8:

Increasing Demand Order Variability Upstream the Supply Chain..............................35

Figure 9:

Logistics Planning Activities ...............................................................................................41

Figure 10: Examples of Hierarchical Situations ..................................................................................43 Figure 11: Structure of a Hierarchical Planning System.....................................................................43 Figure 12: Hierarchies in Distributed Planning...................................................................................45 Figure 13: Hierarchical Structure of Logistics Planning in Supply Chains......................................47 Figure 14: Forms of Inter-Organizational Planning...........................................................................48 Figure 15: Multi-Stop Shipping and Sequenced Loading ..................................................................64 Figure 16: Delivery to the Manufacturer with and without Merge-in-Transit................................65 Figure 17: Comparison of Merge-in-Transit and Cross Docking ....................................................68 Figure 18: Bilateral Information Exchange in Supply Chains...........................................................72 Figure 19: The Supply Chain-Wide Exchange of Information.........................................................72

XVI

List of Figures

Figure 20: Screenshot from ICON-SCC..............................................................................................75 Figure 21: Complementary Use of ERP and SCM Solutions ...........................................................79 Figure 22: Is Your Company Taking Part in Multiple Supply Chains?............................................81 Figure 23: Degree of Collaboration in Planning Processes...............................................................82 Figure 24: Does Your Company Have a SCM Software Solution?..................................................84 Figure 25: Why Does Your Company not Use SCM Software?.......................................................85 Figure 26: Persecuted Goals – Reductions ..........................................................................................86 Figure 27: Persecuted Goals – Improvements ....................................................................................87 Figure 28: Evaluation of Supply Chain Management Software – I..................................................88 Figure 29: Evaluation of Supply Chain Management Software – II ................................................89 Figure 30: Selection Criteria for SCM Software Solutions ................................................................92 Figure 31: Worldwide Locations of Audi Group................................................................................99 Figure 32: Transportation between Audi in Györ and Ingolstadt/Neckarsulm as well as Other Plants............................................................................................................104 Figure 33: The Supply Chain of the V8 4.0l Diesel Engine ............................................................107 Figure 34: Backlogs Quoted in Audi Neckarsulm’s Supply Call-offs to Audi Hungaria between Week 10 and Week 26 of 2003 ........................................................113 Figure 35: Wahler’s Inventory Level of Thermostats between Week 10 and Week 26 of 2003 ............................................................................................................................115 Figure 36: TCG Systemtechnik's Inventory Level of Thermostats between Week 10 and Week 26 of 2003..........................................................................................................116 Figure 37: TCG Systemtechnik’s Inventory Level of Water Pumps between Week 10 and Week 26 of 2003 ....................................................................................................117 Figure 38: Audi Neckarsulm’s Inventory Level of V8 4.0l Diesel Engines between Week 10 and Week 26 of 2003 .........................................................................................119 Figure 39: Ratio of Audi Neckarsulm’s Demand in a Particular Week in the Final Supply Call-off, and Demand in the Same Week in the Supply Call-off Four Weeks Earlier (Calculated between Week 10 and Week 26 of 2003)................126

List of Figures

XVII

Figure 40: Demand Order Variability in the Supply Chain for the V8 4.0l Diesel Engine...................................................................................................................................130 Figure 41: The Architecture of the SCOptimizer .............................................................................162 Figure 42: Excerpt from an XML Description for a Solver Class .................................................163 Figure 43: XML Registry for at Runtime Available Distribution Planning Methods..................164 Figure 44: Excerpt from the XML Registry of Available Actors for Modeling ...........................164 Figure 45: Excerpt from the Description File for Warehouses ......................................................166 Figure 46: Excerpt from the XML Description of a Desk Session ...............................................167 Figure 47: Excerpt from a Results File for Centralized Distribution Planning............................168 Figure 48: Non-Cooperative Transportation Planning....................................................................171 Figure 49: Centralized Cooperative Transportation Planning ........................................................171 Figure 50: Modeling Mask of the SCOptimizer ................................................................................175 Figure 51: Selecting a Planning Method for the Vehicle Routing in the Distribution Planning................................................................................................................................176 Figure 52: Mask for Selecting Cooperation Degree and Cooperating Nodes ..............................177 Figure 53: Input Mask for the Savings Method of Clarke and Wright with Point Graph....................................................................................................................................178 Figure 54: Tabular Display of Results for the Decentralized, Non-Cooperative Planning................................................................................................................................181 Figure 55: Graphic Display of Results for the Decentralized, Non-Cooperative Planning................................................................................................................................182 Figure 56: Tabular Display of Planning Results for the Decentralized, Cooperative Planning................................................................................................................................183 Figure 57: Graphic Display of Results for the Decentralized, Cooperative Planning.................184 Figure 58: Tabular Display of Results for the Centralized, Cooperative Planning ......................185 Figure 59: Graphical Display of Results for the Centralized, Cooperative Planning ..................186 Figure 60: Exemplary Model for the Evaluation of the Bullwhip Effect......................................200 Figure 61: Selecting a Planning Method for the Evaluation of the Bullwhip Effect...................200

XVIII

List of Figures

Figure 62: Excerpt from the Description File for the Bullwhip Effect Evaluation ....................201 Figure 63: Option Dialog with Required Input from the Planner .................................................202 Figure 64: Input Mask for Selecting the Degree of Cooperation...................................................202 Figure 65: Input Mask for Determining the Characteristics of Historic Demand Data.......................................................................................................................................203 Figure 66: Input Mask for Selecting the Corresponding Bill of Materials ....................................203 Figure 67: Bill of Materials for a Single-Path Supply Chain ............................................................204 Figure 68: Excerpt from the Description File for the Input Parameters for the Bullwhip Effect Evaluation ...............................................................................................205 Figure 69: Input Mask for Actor-Dependent Parameters for the Bullwhip Effect Evaluation ............................................................................................................................205 Figure 70: Planner’s View of the Input Table for Costs Involved in the Bullwhip Effect Evaluation................................................................................................................206 Figure 71: Graphic Display of Results of the Bullwhip Effect Evaluation for the Decentralized, Non-Cooperative Scenario .....................................................................208 Figure 72: Tabular Display of Results of the Bullwhip Effect Evaluation for the Decentralized, Non-Cooperative Scenario .....................................................................209 Figure 73: Tabular View of Total Order Amounts in the Decentralized, NonCooperative Scenario..........................................................................................................210 Figure 74: Tabular View of Inventory Costs per Period in the Decentralized, NonCooperative Scenario..........................................................................................................211 Figure 75: Tabular View of Total Inventory Costs in the Decentralized, NonCooperative Scenario..........................................................................................................212 Figure 76: Graphic Display of Inventory Levels in the Decentralized, NonCooperative Scenario..........................................................................................................213 Figure 77: Graphic Display of Results of the Bullwhip Effect Evaluation for the Decentralized, Non-Cooperative Scenario .....................................................................215 Figure 78: Graphic Display of Inventory Levels in the Decentralized, Cooperative Scenario ................................................................................................................................216

List of Figures

XIX

Figure 79: Graphic Display of Results of the Bullwhip Effect Evaluation for the Centralized, Non-Cooperative Scenario..........................................................................218 Figure 80: Graphic Display of Inventory Levels in the Centralized, Cooperative Scenario ................................................................................................................................219 Figure 81: Steps of the Computational Study....................................................................................223 Figure 82: Overview of the Experiment Design of the Computational Study.............................225 Figure 83: Total Supply Chain Costs ..................................................................................................230 Figure 84: Results of the Centralized, Cooperative Experiments in Terms of Total Supply Chain Costs.............................................................................................................230 Figure 85: Results of the Decentralized, Cooperative Experiments in Terms of Total Supply Chain Costs ..................................................................................................231 Figure 86: Results of the Decentralized, Non-Cooperative Experiments in Terms of Total Supply Chain Costs ..................................................................................................231

List of Tables

Table 1:

Streams of Research Contributing to the Field of Supply Chain Management .............................................................................................................................24

Table 2:

Sample Definitions for “Supply Chain”...............................................................................24

Table 3:

Sample Definitions for “Supply Chain Management”.......................................................25

Table 4:

Four Main Uses of the Term “Supply Chain” ....................................................................26

Table 5:

Objectives of Supply Chain Management............................................................................29

Table 6:

Commonly Used Models of Organizational Effectiveness ...............................................31

Table 7:

Decision Categories in Logistics ...........................................................................................40

Table 8:

Examples of Services Offered by Logistics Service Providers .........................................57

Table 9:

Examples of Cooperation between Manufacturers and LSPs ..........................................58

Table 10: Examples of the Implementation of Merge-in-Transit......................................................66 Table 11: Examples of the Implementation of Cross Docking .........................................................69 Table 12: Examples of Extensive Information Exchange between Companies .............................77 Table 13: Dependencies ERP-Supply Chain Management Software Solution................................93 Table 14: Key Figures of Audi Group for 2003 and 2004..................................................................99 Table 15: Production of Vehicles for Audi Group in 2003 and 2004.............................................100 Table 16: Audi AG’s Position in Major Markets................................................................................101 Table 17: Summary of Key Inventory Figures ...................................................................................121

XXII

List of Tables

Table 18: Orders of Audi Neckarsulm, Audi Hungaria, and TCG Systemtechnik between Week 10 and Week 26 of 2003 (Weekly Amounts)..........................................124 Table 19: Quantification of the Bullwhip Effect in the Engine Supply Chain between Week 10 and Week 26 of 2003 ............................................................................130 Table 20: Summary of Evaluation of Cooperation Scenarios ..........................................................155 Table 21: Lead Times between Members of the Engine Supply Chain..........................................156 Table 22: Vehicle Parameters ................................................................................................................179 Table 23: Further Input Parameters.....................................................................................................179 Table 24: Distance between Nodes in the Plane................................................................................180 Table 25: Summary of Supply Chain Results of the Exemplary Evaluation of Distribution Planning............................................................................................................187 Table 26: Individual Performance of Cooperating Partners.............................................................188 Table 27: End Customer’s Demand per Period .................................................................................206 Table 28: Order Amounts of Each Actor for Each Period in the Decentralized, Non-Cooperative Scenario...................................................................................................211 Table 29: Results of the Decentralized, Non-Cooperative Scenario of the Exemplary Bullwhip Effect Evaluation..................................................................................................213 Table 30: Order Amounts of Each Actor for Each Period in the Decentralized, Cooperative Scenario ............................................................................................................215 Table 31: Results of the Decentralized, Cooperative Scenario of the Exemplary Bullwhip Effect Evaluation..................................................................................................217 Table 32: Order Amounts of Each Actor for Each Period in the Centralized, Cooperative Scenario ............................................................................................................218 Table 33: Results of the Centralized, Cooperative Scenario of the Exemplary Bullwhip Effect Evaluation..................................................................................................220 Table 34: Results of the Exemplary Bullwhip Effect Evaluation ....................................................221 Table 35: Key Scenario Parameters for the Creation of Experiments............................................224 Table 36: Vehicle Transfer Costs between DCs (Fixed plus Variable Costs)................................224

List of Tables

XXIII

Table 37: Performance of Experiments (in Percentage) with an External/Internal Vehicle Price Ratio of 0.5.....................................................................................................232 Table 38: Performance of Experiments (in Percentage) with an External/Internal Vehicle Price Ratio of 1.0.....................................................................................................232 Table 39: Performance of Experiments (in Percentage) with an External/Internal Vehicle Price Ratio of 1.5.....................................................................................................232 Table 40: Performance of Experiments (in Percentage) in a Centralized Cooperation ...............233 Table 41: Performance of Experiments (in Percentage) in a Decentralized Cooperation............................................................................................................................233 Table 42: Performance of Experiments (in Percentage) with No Cooperation ............................234 Table 43: Individual Performance of Experiments (in Percentage) with an External/Internal Vehicle Price Ratio of 0.5.....................................................................234 Table 44: Individual Performance of Experiments (in Percentage) with an External/Internal Vehicle Price Ratio of 1.0.....................................................................235 Table 45: Individual Performance of Experiments (in Percentage) with an External/Internal Vehicle Price Ratio of 1.5.....................................................................235 Table 46: Performance of Experiments in Terms of Vehicle Costs (in Percentage) with an External/Internal Vehicle Price Ratio of 0.5 ......................................................237 Table 47: Performance of Experiments in Terms of Vehicle Costs (in Percentage) with an External/Internal Vehicle Price Ratio of 1.0 ......................................................237 Table 48: Performance of Experiments in Terms of Vehicle Costs (in Percentage) with an External/Internal Vehicle Price Ratio of 1.5 ......................................................237 Table 49: Performance of Experiments in Terms of Distance-Dependent Variable Costs for all Settings of External/Internal Vehicle Price Ratio......................................238 Table 50: Performance of Experiments in Terms of Total Tours for All Settings of External/Internal Vehicle Price Ratio................................................................................239 Table 51: Performance of Experiments in Terms of Distance per Tour for All Settings of External/Internal Vehicle Price Ratio ............................................................239 Table 52: Performance of Experiments in Terms of Retailers per Tour for All Settings of External/Internal Vehicle Price Ratio ............................................................240

XXIV

List of Tables

Table 53: Performance of Experiments in Terms of Vehicle Capacity Utilization (in Percent) for All Settings of External/Internal Vehicle Price Ratio ...............................241 Table 54: Performance of Experiments in Terms of Unused Internal Vehicles with a Retailer Demand/Transportation Capacity Ratio of 0.5 .................................................241 Table 55: Performance of Experiments in Terms of Unused Internal Vehicles with a Retailer Demand/Transportation Capacity Ratio of 1.0 .................................................242 Table 56: Performance of Experiments in Terms of Unused Internal Vehicles with a Retailer Demand/Transportation Capacity Ratio of 1.5 .................................................242 Table 57: Performance of Experiments in Terms of LSP Vehicles Retrieved with a Retailer Demand/Transportation Capacity Ratio of 0.5 .................................................243 Table 58: Performance of Experiments in Terms of LSP Vehicles Retrieved with a Retailer Demand/Transportation Capacity Ratio of 1.0 .................................................243 Table 59: Performance of Experiments in Terms of LSP Vehicles Retrieved with a Retailer Demand/Transportation Capacity Ratio of 1.5 .................................................243 Table 60: Individual Performance of Experiments in Terms of Total Costs per Customer (in Percentage) with an External/Internal Vehicle Price Ratio of 0.5.............................................................................................................................................245 Table 61: Individual Performance of Experiments in Terms of Total Costs per Customer (in Percentage) with an External/Internal Vehicle Price Ratio of 1.0.............................................................................................................................................245 Table 62: Individual Performance of Experiments in Terms of Total Costs per Customer (in Percentage) with an External/Internal Vehicle Price Ratio of 1.5.............................................................................................................................................246 Table 63: Performance of Experiments in Terms of Distance per Retailer for All Settings of External/Internal Vehicle Price Ratio ............................................................246 Table 64: Performance of Experiments in Terms of Total Costs per Tour with a Retailer Demand/Transportation Capacity Ratio of 0.5 .................................................247 Table 65: Performance of Experiments in Terms of Total Costs per Tour with a Retailer Demand/Transportation Capacity Ratio of 1.0 .................................................247 Table 66: Performance of Experiments in Terms of Total Costs per Tour with a Retailer Demand/Transportation Capacity Ratio of 1.5 .................................................247

List of Tables

XXV

Table 67: Individual Performance of Experiments in Terms of Potential Revenue Units with a Retailer Demand/Transportation Capacity Ratio of 0.5...........................248 Table 68: Individual Performance of Experiments in Terms of Potential Revenue Units with a Retailer Demand/Transportation Capacity Ratio of 1.0...........................249 Table 69: Individual Performance of Experiments in Terms of Potential Revenue Units with a Retailer Demand/Transportation Capacity Ratio of 1.5...........................249

Abbreviations

Audi

Audi AG

Audi Hungaria

Audi Hungaria Motor Kft.

Audi Ingolstadt

Audi AG, plant in Ingolstadt

Audi Neckarsulm

Audi AG, plant in Neckarsulm

CAD

Computer aided drawing, also computer aided design

CAE

Computer aided engineering

CAM

Computer aided manufacturing

CEN-COOP

Centralized, cooperative planning

CKD

Completely knocked down

CMI

Co-managed inventory

COFC

Container on a flatcar

CPU

Central processing unit

CRP

Continuous replenishment

DC

Distribution center

DCI

Distribution Centers, Inc.

DEC-COOP

Decentralized, cooperative planning

DMU

Decision-Making Unit

DRAM

Dynamic random access memory

EDI

Electronic data interchange

EDLP

Every-day low price

FAB

Feinabruf (precise supply call-off)

GEKO

Ganzheitliche Ertragsorientierte Kettenoptimierung

GIF

Gewerbe- und Industriepark Friedrichshall

GVZ

Güterverteilzentrum

XXVIII

Abbreviations

ICON

ICON Gesellschaft für Supply-Chain-Management mbH

ICT

Information and Communication Technology

JIT

Just-in-time

LAB

Lieferabruf (supply call-off)

LAFES

Lieferabrufverteilungs- und Feinsteuerungssystem

LCL

Lower control limit

LSP

Logistics Service Provider

LTL

Less-than-truckload

MABES

Materialfluss-Bewegungsverarbeitungs- und Bestandsführungs-System

MIS

Management information systems

MIT

Merge-in-transit

MRP I

Material requirements planning

MRP II

Manufacturing resource planning

NON-COOP

Decentralized, non-cooperative planning

NUMMI

New United Motors Manufacturing Incorporated

OEM

Original equipment manufacturer

P

Punishment

P&G

The Procter & Gamble Company

PD

Pumpe/Diesel

PO

Purchase order

POS

Point of sale

R

Reward

R&D

Research and development

ROP

Reorder point

S

Sucker

SCC

Supply Chain Collaboration

SCE

Supply Chain Execution

Schachinger

Schachinger Logistik Holding GmbH & Co KG

SCM

Supply Chain Management

SCMo

Supply Chain Monitoring

SCP

Supply Chain Planning

SKU

Stock keeping unit

SSL

Secure socket layer

T

Temptation

TCE

Transaction cost economics

Abbreviations

XXIX

TCG Group

Trident Components Group

TCG Systemtechnik

TCG Unitech Systemtechnik

TDI

Turbocharged direct injection

TL

Truckload

TOFC

Trailer on a flatcar

UCL

Upper control limit

US

United States

VMI

Vendor managed inventory

VW

Volkswagen

VW Group

Volkswagen Group

VW T

VW Transport GmbH & Co. OHG

Wahler

Gustav Wahler GmbH u. Co. KG

Wal-Mart

Wal-Mart Stores, Inc.

WIP

Work-in-progress

1 Introduction

Traditionally, competition has been understood as the battle of stand-alone companies for competitive advantage. In order to gain such an advantage, it was believed that “a firm must deliver value to its customers through performing […] activities more efficiently than its competitors or by performing the activities in a unique way that creates greater differentiation” (Christopher, 1998a, p. 11). The first approach is referred to by Porter as cost leadership strategy and the latter as differentiation strategy (Porter, 1985). This approach might be called the “firmcentric model of competition” because it focuses on the rivalry between independent companies for a competitive edge. In addition to this kind of battle which is carried out between entities on the same level of the supply chain, this model includes another kind of battle in which a company on one level of the supply chain struggles with a company on the preceding level for the best price-value ratio [while the preceding company (i.e. the supplier) wants to receive the highest possible price for a given quality or to deliver minimum quality for a given price, and the subsequent company (i.e. the customer) aims at the opposite (i.e. the best quality at the lowest price)]. However, this model might be myopic because it does not take into account the competitive reality of a world characterized by division of labor: Hardly any goods are produced and sold by one company alone. In other words, there is most likely no fully vertically integrated company mastering all alone the entire creation of value starting with the extraction of raw materials and ending with the sale of the end product to the end customer. Therefore, companies in the supply chain “are dependent upon each other by definition and yet paradoxically by tradition do not closely co-operate with each other” (Christopher, 1998a, p. 15), but consider each other as opponents. Since companies obviously depend on each other in order to create value, a more appropriate model might be the “supply chain-centric model of competition” (Christopher, 1998, p. 272).

2

1. Introduction

Here, the focus is on supply chains contending against each other whereby a supply chain can pursue either one of Porter’s strategies. Companies that are members of the same supply chain do not compete, but instead try to cooperate whenever possible and be beneficial to themselves and secondarily, to all1, in order to gain a competitive edge over other supply chains. This new model of competition embraces the idea of inter-network (i.e. inter-supply chain) competition: “Business management has entered the era of inter-network competition. In this emerging competitive environment, the ultimate success of the single business will depend on management’s ability to integrate the company’s intricate network of business relationships” (Bowersox, 1997, as cited in Lambert, 2002, p. 1).

The acknowledgement that a network of cooperating companies, e.g. a supply chain, could be more successful in achieving competitive advantage than individual businesses, constitutes “one of the most significant paradigm shifts in modern business management” (Lambert, 2002, p. 1) as it leaves behind the notion of adversarial companies engaging in fierce competition with one another in order to gain competitive advantage. “In short, the contention that it is supply chains, and not single firms, that compete is a central tenet in the field of supply chain management (Christopher, 1992)” (Macbeth & Ferguson, 1994, as cited in Croom et al., 2000, p. 68). Strictly speaking, the supply chain is not a chain but rather a network (Christopher, 1998, p. 273; Christopher, 1998a, p. 18; Buxmann et al., 2004, p. 5; Pfohl, 2004), and the emerging competitive paradigm suggests that “We are now entering the era of ‘network competition’ where the prizes will go to those organizations who can better structure, coordinate and manage the relationships with their partners in a network committed to better, faster and closer relationships.” (Christopher, 1998, p. 272).

This paradigm shift is also motivated by the gradual change of the markets which in recent years have evolved from an offer-driven to a demand-driven context in which companies have to survive. This evolution of the markets has allowed new characteristics to arise which companies have to deal with: •

Globalization: Firms are in an international context of competition.

•

Demand expectations: Customers’ expectations drive the levels of quality and service in these markets.

1

Ideally, members of a supply chain would also cooperate with other members even if it is not beneficial to themselves, but to the entire system. However, this seems quite unrealistic unless some redistribution system has been installed, or the business pressure of the system towards the individual company is great enough.

1. Introduction •

3

Information and communication technology (ICT): Firms have opportunities of dealing with larger dimensions of complexity and data as well as the integration with business partners.

It is this availability of relevant process information that creates the need for fast reaction times and flexibility, social and institutional orientation, and an accurate management of financial flows (Pfohl et al., 2004) and sets new strategic demands to the way business is done (Bölzing, 2004, pp. 109ff). Thus, strategic concepts such as Supply Chain Management (SCM) have moved to the focus of modern Value Chain Management. Cooperation between partners in the supply chain becomes a key factor to ensure long-term growth (Christopher, 1998, p. 284). The automotive industry is a good example of an industry sector that has to deal with these characteristics of a modern market: Globalization, demand expectations, and advanced ICT (for example, Audi AG has to coordinate the sale of more than two thirds of all vehicle units outside Germany from 10 locations placed around the globe; see also Section 4.1.1.1). Under this prospect, it is worthwhile to ask how cooperative business partners are within supply chains of the automotive industry. Is a supply chain of the European automotive industry an end-to-end cooperative environment, and if not, why not? One significant hint for the adoption of SCM principles in this sector of industry might be the expenses incurred by the players of the automotive industry in SCM software. AMR Research reveals that Automotive and Auto Parts was one of the top four branches in SCM license expenses in 2002 (the others were Computer and Electronics, Food and Beverages, and Retail Trade; Lapide & Davis, 2003, p. 24). In addition, the decreasing manufacturing penetration of car manufacturers (McIvor et al., 1998; Mathisson-Öjmertz & Johansson, 2000; AI, 2002; Kinkel & Lay, 2003, p. 5; AI, 2005) leads to corresponding requirements regarding SCM and SCM software solutions. It is the purpose of this dissertation to present an insight into how cooperation is achieved in supply chains of the European automotive industry and to explore how cooperation can be reinforced by addressing the main challenges companies have to face when defining a cooperative strategy at a planning level. This dissertation analyzes both the possibilities for cooperation in the context of transportation and the bullwhip effect as an indicator for the quality of cooperation in supply chains. In addition, it explores the use of SCM software as an indicator for adopting SCM principles. Furthermore, this dissertation examines the main challenges in creating a cooperation network among supply chain partners. It also describes the implementation of a software prototype for aiding the decision making process in cooperative strategies.

4

1. Introduction

1.1 Motivation and Research Questions Some press releases and interviews conducted by the author with representatives from the automotive industry allowed the sector to shift to the prospect in which the company-centric model of competition seems to still best describe the situation within this branch. It is also believed to be the cause of extended costs and quality deficits in vehicles from German manufacturers which have, in the past, predominantly been state of the art in quality measures (Wassink, 2004). These manufacturers are facing increased customer dissatisfaction, the source of which is believed to lie in the competitive relationship between them and their suppliers (ADAC, 2004; N-TV, 2003; SPIEGEL, 2003; Manager-Magazin, 2004, Manager-Magazin, 2003). In their attempt to lower purchasing costs, which are designed to lower the total production costs and, thus, appear more competitive to the end customer, some manufacturers (namely Mercedes Benz, a brand by DaimlerChrysler) are going so far as to require 150 suppliers to lower the price for parts by 5 percent per year for the next three years. These suppliers are to accomplish an additional price reduction of five percent for already supplied parts. The fact that the cost pressure is increasingly being passed on to the suppliers creates further risks to the manufacturers who eventually have to face the withdrawal of suppliers and so face shortages (Wassink, 2004). A measure against these risks, and a measure to maintain competitiveness is increased collaboration with partners rather than a shift in cost pressure (Holmes & Darivemula, 2004, pp. 2f). On the other hand, there is still “significant room for improvement in collaboration between suppliers and vehicle manufacturers in areas such as cost reduction/cost effectiveness and product development” (Holmes & Darivemula, 2004, p. 5). These facts, in addition to the interviews led by the author with representatives from companies of the European automotive industry, who pointed out the lack of established cooperation between partners in the supply chain, lead to the central research questions of this work: •

What are the causes for the reluctance to establish an end-to-end cooperative environment in supply chains?

•

If cooperation is implemented, what are the causes for the inefficiencies (if any) in its execution?

•

How can these causes be addressed in order to avail cooperation within the supply chain (at the core of supply-chain-centric competition) as the better strategy rather than company-centric competition?

The goals implied in these research questions are:

1.1 Motivation and Research Questions •

5

Defining the scope of cooperation within the context of SCM. This requires the definition of both theories: Theory of Cooperation and Theory of Supply Chain Management.

•

Identifying ways of cooperation in the context of SCM and selecting exemplary cooperation scenarios to define the scope of this dissertation.

•

Providing empirical evidence (in the form of a case study; Benbasat et al., 1987; Lee, 1989; Yin, 2002; Dubé & Paré, 2003) that a lack of cooperation leads to increased costs and financial disadvantages in supply chains.

•

Empirically exploring the causes for the reluctance to implement an end-to-end cooperative environment in supply chains of an exemplary industry.

•

Implementing an application prototype addressing the main cause for non-cooperation and for aiding decisions on cooperative strategies among partners in the supply chain.

•

Performing an exemplary computational study to prove the applicability of the prototypic implementation and derive more general statements on the benefits of cooperation.

The prototypic implementation will focus on the evaluation of different cooperative strategies in the planning process of selected supply-chain-related scenarios. For this purpose, it is imperative to find a systematic method to assess the value of cooperative logistics planning. The sub-goals implied herein are: •

Identification of an operative method of categorizing conceptual degrees of cooperative planning in supply chains. This implies a combination of degrees of cooperation and selected planning scenarios.

•

Design of an architecture for comparing and evaluating different ways of cooperative planning in supply chains.

•

Prototypic implementation of the evaluation architecture.

•

Selection of exemplary logistics planning activities for comparison.

•

Proof-of-concept execution of a comparative computational study on a selected exemplary logistics planning activity.

Furthermore, from a more generic point of view, this dissertation is to present an insight into the potential of cooperation in supply chains. For this purpose, scenarios will be presented that describe how members of a supply chain could cooperate with each other in the field of SCM, more specifically in the field of transportation, and it will be evaluated if they could thereby gain an advantage in terms of improving service levels or reducing costs (e.g. by alleviating the

6

1. Introduction

bullwhip effect) and, thus, reducing its negative impacts on the supply chain as a whole as well as on its individual members.

1.2 Structure of the Dissertation First, the author presents research and the underlying theories in the fields of cooperation, logistics, and SCM (Chapter 2). In this section, the semantics of the key terms of this dissertation are defined. It delivers the basics for the next part of the dissertation in which the field of cooperation of a selected area of SCM are described. Namely, transportation policy is a field of cooperation for which Chapter 3 presents examples of cooperative behavior taken from research literature. Chapter 3 closes with an empirical study that aims to show how cooperative supply chains of the European automotive industry are and how SCM software is used as a cooperative instrument. Since the theory of SCM praises the advantages of cooperation, and literature provides a variety of examples of how cooperation can be put into practice, this survey aims to explore the main drawbacks of cooperation within logistics activities in the European automotive industry. This survey reveals among other indicators that, for the sample of firms included, the main challenge for cooperation with supply chain partners is the difficulty of assessing the value of the cooperation. The follow-up question arises: How can the benefits of cooperation be quantified? That the lack of cooperation or also a wrong design of cooperation can lead to added costs in the supply chain is shown in Chapter 4. This section describes a specific supply chain of Audi AG and shows that despite the existence of cooperation agreements, there still is an accentuated bullwhip effect causing a surplus of demand variability. The chain of the Audi A8 V8 4.0l diesel engine2 is evaluated in order to analyze to which extent a cooperation scenario might be beneficial. When assessing the advantages of a scenario, the focus will be on a scenario’s potential for mitigating the negative impacts of the bullwhip effect on the supply chain as a whole and on its individual members whereby these adverse impacts could be reduced if the effect itself is lessened.

2

Between March and June 2003 there were actually two different types of this engine. The difference depends on whether the steering wheel is placed on the left side, or on the right side. At Audi, these engine variations are referred to as “Motorsorte 8450 W18 Rechtslenker” and “Motorsorte 8499 W18 Linkslenker”. When referring to the V8 4.0l diesel engine, both types are meant. As can be seen in BESI2 (c.f. Appendix J), these two engines were discontinued in week 34 of 2003 and were replaced by another type of engine referred to as “Motorsorte 8451 W18 Links- und Rechtslenker” (part number: 057100015D), which was produced in series starting in week 34 of 2003 (information extracted from BESI2, June 26, 2003). BESI2 is an information system which determines gross demand for a certain part number on a weekly basis (e.g. 057100015D for the upcoming kind of V8 diesel engine) based on forecast data and the bill of materials. The result of each calculation run is the weekly gross demand for the upcoming four months and the monthly demand for the two months after that. Gross demand is calculated weekly by BESI2 and the calculation of the current week always supersedes the calculation from the previous week (Schmidt, 2002).

1.2 Structure of the Dissertation

7

Chapter 5 presents an application prototype as an auxiliary means (König, 1994, p. 81) that uses state-of-the-art technology to combine conclusions from the Theory of Cooperation (namely, the structuring of categories for distributed decision finding as described by Wyner and Malone (1996) and the categorization of Schneeweiss (1999, p. 5)) with planning methods from the field of logistics to open up an operative way of quantifying the benefits of cooperation in supply chains. This prototype allows one to model participants of the supply chain with their respective information and product flows. Consequently, the tool provides an architecture for implementing planning scenarios (e.g. distribution planning) which can be run with different degrees of cooperation. The comparison of the results of different cooperation levels on a single planning scenario allows one to assess the benefits of different cooperation strategies (of course, most simulation scenarios are models of reality, implying a simplification that puts their extrapolation into perspective. This is the case for the examples given in this work). This chapter also describes a computational study on cooperative distribution, which shows how the variation of different parameters affect the quality of the results and thus the added value of cooperation. The dissertation will close with Chapter 6, which will provide a summary of the findings. Based on these findings, implications will be drawn on the advantages of cooperation in SCM. Furthermore, issues encountered in this research will be pointed out. Last but not least, some final thoughts on SCM will be stated and a recommendation for future research will be given.3

3

The reader is kindly requested to note that all references to information enclosed in the Appendix of this dissertation can be found online at http://www.martindiaz.net/dissertation

2 Inter-Organizational Cooperation and Supply Chain Management

For many terms used in this dissertation no commonly accepted definitions exist. The understanding of the terms might be different from one research discipline to another which makes a validity of its use difficult. For the context of this dissertation, some terms might be subject of discussion when it comes to finding a universal definition. In fact, it is even problematic to find a common definition in the field of economics. Because of this and in order to sharpen the meaning of the statements formulated in this work, the following terms will be understood as follows: •

Collaboration. Collaboration is used in this dissertation as the act of working together, irrespective of whether this is done explicitly or not, voluntarily or not. Collaborating parts (note: always more than one part) commonly perform tasks and jointly solve problems.

•

Coordination: To coordinate in this dissertation is understood as “to manage dependencies among activities so as to achieve coherent operation of the entire system in question.” (Teigen, 1997)

•

Competition: The act of competing, as for profit or a prize; rivalry between two or more businesses striving for the same customer or market (AHD, 2000).

•

Planning: In light of the business focus of this dissertation, planning is understood to be the identification, definition, and commitment to future activities intended to achieve the goals of the company (Fleischmann, 2002, p. A1-9).

Tröndle (1987, p. 13) finds it difficult to define a term when this one term is commonly used in non-formal language and in scientific language as well as in the context of single research fields and in an inter-disciplinary context. This clearly applies to the terms mentioned above which are

10

2. Inter-Organizational Cooperation and Supply Chain Management

key elements of formulations and descriptive as well as normative statements within this dissertation. This also applies to the terms cooperation and co-opetition. These two are subject to more explicit discussion in Sections 2.1.1 and 2.2.4.1. This dissertation focuses on cooperative planning between firms in the context of a supply chain. This planning concerns a number of players – namely suppliers, manufacturers, logistics service providers, carriers, wholesalers, and retailers – who are involved in the processes of creating value in the supply chain, in form of products and/or services addressed to the consumer (Christopher, 1998a, p. 15; Stadtler, 2000, p. 7; Buxmann et al., 2004, p. 2). The management of these processes is what is called SCM in the literature. As SCM is a contextualization of inter-organizational cooperation, this chapter begins with an overview of the theoretical aspects of inter-organizational cooperation (Section 2.1). This section first defines the term cooperation (Section 2.1.1) and the term inter-organizational cooperation (Section 2.1.2). It also presents a method of categorizing the types of cooperation (Section 2.1.3). Thereafter, the fundamentals of SCM are presented (Section 2.2). This section gives an overview of the definitions for supply chain and SCM and describes the business processes involved in SCM (Sections 2.2.1 and 2.2.2). Section 2.2.3 describes the goals pursued by this strategic concept. Section 2.2.4 describes the main issues related to cooperation in the context of SCM such as co-opetition (Section 2.2.4.1). In addition, the so-called “bullwhip effect” is described as an example of how cooperation can analytically be proven to be advantageous in the very concrete context of SCM (Section 2.2.4.2). This work focuses on the logistic aspects of SCM. Section 2.3 describes logistics planning as an object of inter-organizational cooperation. The fields of decision making in logistics and the planning activities are described first (Section 2.3.1). Then, logistics planning is shown to be a hierarchical planning problem (Section 2.3.2), and analogously, inter-organizational logistics planning is shown to be a multi-tiered hierarchical problem (Section 2.3.3). Following this contextualization of inter-organizational planning, Section 2.3.4 presents the approach of Wyner and Malone (1996) which serves as the basis for comparing different degrees of cooperation in the remainder of this dissertation.

2.1 Inter-Organizational Cooperation SCM is concerned with managing the flow of information as well as of materials and products between members of a supply chain (see Section 2.2). To realize a smooth and speedy flow, it is necessary that all involved parties work together closely, i.e. that they cooperate.

2.1 Inter-Organizational Cooperation

11

2.1.1 Definition of Cooperation The term “cooperation” was first described in 1394 and has its etymological origin in the Latin “cooperationem,” a noun formed from “co” –with– and “operari” –to work–, which comes from the noun “opus” –work– whose genitive is “operari” (see http://www.etymonline .com/c9etym.htm). The Merriam Webster Dictionary (MWD, 2004) defines cooperation as the “association of persons for common benefit.” The American Heritage Dictionary (AHD, 2000a) describes it as the “association of persons or businesses for common, usually economic, benefit.” Considering that the term “cooperation” is very important, it still remains under a fuzzy shape because and despite of its widespread use (Emmerich, 2001, pp. 92f). Even in single scientific disciplines, it has not been possible to form a sharp definition of the term. And this is a tragic aspect in Cooperation Theory because research can only be based on minimal agreement of basic terminology (Grunwald, 1982, p. 73). One approach to define term-constitutive characteristics of cooperation from an interdisciplinary point of view is presented by Tröndle (1987, pp. 16f). He comes to two “constitutive dimensions” which are weighted differently in several fields of research:4 Autonomy and Interdependence. Tröndle describes these dimensions of “cooperation” as follows: “The apparent contradiction between Autonomy and Interdependence is explained by the fact that the participating cooperation partners have the ultimate power to decide if joining or leaving, without underlying the directives from a superior authority. Thus, they are autonomous and are in a relationship among equals.” (Tröndle, 1987, p. 16)5

On the other hand, dependencies exist between the partners in this relationship “because it is essential for the cooperation to settle agreements, this is to commonly make decisions, in order to coordinate behaviors ex ante” (Tröndle, 1987, p. 16).6 Boettcher (1974, p. 42) defines this relationship between the participating partners as the “Cooperation Paradox”7 which for Sölter (1966, p. 6) leads to the “ancient antinomy between freedom and attachment.” Leaving the inter-disciplinary point of view, in the research field of economics there seems to prevail a general understanding of cooperation as collaboration between – not within – firms

4

Refer for example to Marwell & Schmitt, 1975, p. 5; Plaßmann, 1974, pp. 9f; Schwarz, 1979.

5

Citation translated from German by the author.

6

Citation translated from German by the author.

7

Also refer to Grunwald, 1981, p. 74.

12

2. Inter-Organizational Cooperation and Supply Chain Management

(Blohm, 1973, pp. 890f).8 But again, a closer look at the literature reveals that cooperation may have many faces, depending on the focus of attention set by the author (see, for example, Benisch, 1973, p.67; Gerth, 1971, p. 17; Schwarz, 1979, p. 85; Boehme, 1986, p. 24; Harms, 1973, pp.9f; Knoblich, 1969, p. 497; Sölter, 1966a, p. 236). In order to develop a definition of cooperation, Plaßmann (1974, pp. 11f) and Rotering (1993, pp. 8f) both propose to first delimit the term “cooperation” and, in a second step, to define the characteristics of the term. In this manner, cooperation can be understood as an alternative coordination form between “Market” and “Hierarchy” (Rotering, 1993, p. 8).9

Cooperation

Limit

Market

Limit

collaboration is conscious and arranged explicitly

Hierarchy

collaboration may anytime be dismissed by one side

Figure 1: Delimitation of the Coordination Forms Market-Cooperation-Hierarchy (Source: Based on Rotering, 1993, p. 14)

Both terms, Market and Hierarchy, as end points of a coordination continuum find their roots in the Transaction Cost Theory (Rotering, 1993, p. 9).10 As Figure 1 shows, there is a limit between activities coordinated by the nature of the market and activities coordinated by a cooperation agreement (see the “Limit” on the left-hand side of Figure 1). On the other hand, there is also a border that separates activities coordinated by cooperative behavior and activities ruled by hierarchical relationships (see the “Limit” on the right-hand side of Figure 1).

8

For cooperation within the firm refer, for example, to Zörgiebel, 1983, pp. 102f; Schwarz, 1979, pp. 88f; Endress, 1991, pp. 74f.

9

In the context of this dissertation, the term Hierarchy can be derived into the terms group, corporation, merger, or acquisition (Rotering, 1993, p. 9). Refer also to Büchs, 1991, pp. 3f.

10

Other theoretical frameworks that allow delimiting “cooperation” are, for example, the Theory of Organizational Equilibrium (Barnard, 1968; see also Simon, 1947; March & Simon, 1993; Boettcher, 1974) and the Contingency Theory as a reaction to Weber (1922) and the older work on Organization Theory (Pugh & Hickson, 1971; Pugh, 1981; Frese, 2000, pp. 313f; Kieser & Walgenbach, 2003). Refer to Rotering (1993, pp. 66-147) for an evaluation of these approaches for delimiting “cooperation”.

2.1 Inter-Organizational Cooperation

13

Concerning the border between market-driven and cooperation-oriented coordination, research offers a range of differentiated aspects that characterize the transition from one to other. One requirement mentioned for cooperation is the spin-off or merger of single functions or tasks of the company which are then to be solved collaborative (Benisch, 1973, p. 67; Gerth, 1971, p. 17). This criterion is, though, not a constitutive feature of the term cooperation since, for example, a joint venture between two car manufacturers in order to produce a new car type which they were yet not producing would not represent cooperation (Rotering, 1993, p. 9). In order to split or join functional areas, the collaborating companies must have been performing these task autonomously before which is not the case in the above example. In addition, a spontaneous information exchange between companies would also not be understood as cooperation if one were to follow the above criterion. However, both, joint ventures and mere information exchange, have to be counted as cooperation (Plaßmann, 1974, p. 17). Other references speak of the legal contract as a condition for cooperation (see, for example, Bidlingmaier et al., 1967, p. 353). In the context of this dissertation, this criterion will be interpreted as the explicit arrangement of collaboration without going deeper into the legal aspects of contracting. Thus, a cartel, in which the participants agree on guidelines of behavior, is to be understood as cooperation, too (Plaßmann, 1974, pp. 20f; Rotering, 1993, p. 10). On the other hand, the mere adjustment of conduct (as, for example, in a “peaceful oligopoly”) is not to be understood as cooperation in the context of this dissertation (Blohm, 1980, p. 890; see also Plaßmann, 1974, p. 17 for an opposite opinion). Boettcher (1974, p. 22) differentiates between conscious and unconscious cooperation and concludes that it might be better to understand the latter as a market-driven consequence (see also Eschenburg, 1971, p. 4).11 Other conditions mentioned for delimiting cooperation of market activity are, for example, longterm orientation (Meimberg, 1970, p. 245), organizational autonomy (Schwarz, 1979, p. 85), or the legality of the collaboration (Sölter, 1966a, p. 236). The first two aspects do not apply to clear cooperative activities like, for example, the exchange of sales data between firms for better forecasting accuracy (Simchi-Levi et al., 2002, pp. 109ff; Rojek, 2004). This cooperation does not require to be long-term oriented (it might be used for rolling forecasts, for example, weekly) nor to take place in the context of a newly founded organization. Regarding the legality of the collaboration, it can be said that it is not a term-constitutive feature since some cooperation activities that are illegal in some countries are, therefore, not to be considered as non-cooperative in the country it takes place in. 11

For the origins of the differentiation between conscious and unconscious cooperation refer to Mill, 1924, pp. 174f.

14

2. Inter-Organizational Cooperation and Supply Chain Management

Concerning the border between cooperation-driven and hierarchical coordination, the literature seems to agree on the autonomy of the collaborating partners as being the differentiating aspect (Benisch, 1972, p. 152; Rotering, 1993, p. 11). In this context, autonomy is to be understood as both the legal and economic independence of the collaborating partners (Plaßmann, 1974, p. 13; Tröndle, 1987, p. 16). Thus, the following definition for “cooperation” can be proposed (Rotering, 1993, p. 13; Knoppe, 1997, p. 35): Cooperation is the conscious, explicitly arranged collaboration that can be dismissed anytime by one collaborating partner. However, the fact that cooperating partners remain autonomous and the cooperation is voluntary opens a series of aspects that have to be taken into consideration and greatly enhance the complexity of the Theory of Cooperation. Further implications related to the Game Theory and also socio-cultural aspects involved in the interaction between individuals are fields that are not to be considered any further in this dissertation but remain subject of major research efforts.12

2.1.2 Definition of Inter-Organizational Cooperation Today’s supplier-manufacturer relationships are often partnership-like rather than arm’s length or even adversarial. In other words, today’s relationships are often more cooperative than competitive – at least, that is what the literature could make one believe. Before beginning an exploration of how members of a supply chain could possibly cooperate, it is necessary to first clarify the meaning of “inter-organizational cooperation.” According to Bensaou (1997, p. 109), this term can be defined “as the degree to which focal activities to the relationship are carried out jointly.” The author continues by describing the nature of inter-organizational cooperation: “Implicit in this definition is ‘the interpenetration of organizational boundaries’ (Heide and John, 1990, p. 25; see also Guetzkow, 1966) which implies more than just the sequential division of labor and tasks conducted within a cooperative climate. In the traditional competitive model the responsibilities for key tasks are allocated along a clear division of labor and a strong relational asymmetry in terms of ownership of a product or rents appropriation. In partnership-like relationships cooperation can occur over a large set of activities, including long range planning, development and product design, quality and delivery coordination, training, technical assistance and education” (Bensaou, 1997, p. 109).

12

For these topics refer, for example, to Schelling, 1963; Kenneth, 1986; Gowa, 1986; Jervis, 1988; Milner, 1992; Axelrod, 1997; Lake & Powell, 1999; Axelrod, 2000;

2.1 Inter-Organizational Cooperation

15

Arrighetti et al. (1997) include some further characteristics of inter-organizational cooperation13 in their definition: “[T]he concept of inter-firm cooperation refers, more precisely, to relations of mutual dependence between organizations which retain their separate identity as legal and/or economic entities. […] Inter-firm cooperation implies a long-term orientation of some kind which is common to both parties. […] Inter-firm cooperation in the sense used here also implies a degree of risksharing and exchange of information and expertise during performance, which may take such forms as the joint ownership or cross-ownership of patents and machines, the joint development of products and processes, the shared development of marketing strategies, or the exchange of staff” (Arrighetti et al., 1997, p. 172).

As implied by this definition that risk sharing might be one reason for engaging in a cooperative relationship with other companies. Jarillo and Stevenson (1991, p. 67) point out another motive: They argue that companies choose to cooperate with each other instead of having an arm’s length or adversarial relationship because it enables each participating party to focus on its core competencies; that is, on what it can do best, and to take advantage of the other companies’ superior skills for all other activities.14 In some situations, it might be difficult to determine whether a relationship between a supplier and a customer of a product or service is a cooperative one, or a (long-term) market relationship which might even be adversarial in nature. Notwithstanding the importance of determining the nature of a relationship, this issue will not be addressed any further because – in the opinion of the author – it does not affect the nature of the fact that cooperation is taking place as the following explanation shows. Besides descriptive characteristics of the term “cooperation”, the literature often offers normative features that include statements concerning the objectives of cooperation (see the definitions above and also, for example, Benisch, 1973, p. 67).15 Linné (1993, pp. 24f) proposes a three-way differentiation of cooperation:

13

They use the synonym “inter-firm cooperation.”

14

A core competency “provides potential access to a wide variety of markets […], should make a significant contribution to the perceived customer benefits of the end product […], [and] should be difficult for competitors to imitate” (Prahalad & Hamel, 1990, p. 83). For more on the concept of core competencies, see Prahalad and Hamel (1990), who developed the main ideas of this concept, and Grant (1998) for a more general discussion of a firm’s competencies and core competencies.

15

For a differentiation between descriptive and normative characteristics of the term “cooperation” refer to Salje, 1981, pp. 3f.

16

2. Inter-Organizational Cooperation and Supply Chain Management

1. Cooperation as an instrument. Cooperation which aims to coordinate tasks in order to increase performance helps to lower costs or to reduce time and risk. 2. Cooperation as an institution. Cooperation represents an institution of a certain type in a social system. It describes the structure of the various decision-making instances of a company in relation to collaboration and each of their surrounding systems. 3. Cooperation as a process. Cooperation describes the phases of the collaboration from the initiation across the selection of partners, constitution, management, through its termination. However, the inclusion of normative characteristics in the definition of cooperation is not of relevance for this dissertation due to two main reasons (Rotering, 1993, pp. 12f). First, expressions like “improvement of competitiveness” and similar wordings are not cooperationspecific objectives and are, therefore, not needed for the definition. Second, the inclusion of normative characteristics of the term often assumes that the collaborating partners are seeking the same goals which is not always the case (Plaßmann, 1974, p. 22 and pp. 41f). Thus, “cooperation” can primarily be defined with the use of descriptive characteristics; normative components merely have a complementary nature. Due to these considerations, Rotering (1993, p. 13) defines inter-organizational cooperation as the conscious, explicitly arranged collaboration between organizations that can be dismissed anytime by one of the collaborating organizations. This definition will be the basis for the further discussion in this dissertation.

2.1.3 Forms of Inter-Organizational Cooperation The form of cooperation is often described to be decisive for the success of the collaboration (see, for example, Timmermann, 1985, p. 212). In order to focus the discussion of this dissertation, it is first necessary to systematize the forms of cooperation. This can be done either by enumerating existing forms of cooperation in industry or by differentiating and enhancing the constitutive characteristics of the term cooperation. Based on the above definitions of inter-organizational cooperation, the following criteria could be used to identify the type of relationship: •

Long-term orientation,

•

working towards a common goal,

•

jointly carrying out important activities such as R&D, training, planning,

•

joint ownership of equipment, patents, etc.,

2.1 Inter-Organizational Cooperation •

exchange of personnel, expertise, or information,

•

sharing of risks and benefits, and

•

mutual dependence.

17

Cases in which a relationship matches all of these characteristics are probably rare. However, even if some of these requirements are not satisfied, a relationship can nevertheless be a partnership. To reach a decision as to whether a relationship is a partnership or a market relationship is often discretionary. Henzel (1968, pp. 797f), for instance, presents three characteristics for structuring forms of cooperation: Sort and duration, direction, and object of cooperation. He combines them with a set of properties (for example, for the direction of cooperation: Horizontal, forward vertical, and backwards vertical) and mentions a number of examples. However, these criteria and their properties cannot be fully combined which would be necessary for a systematization (Henzel, 1968, p. 804). Furthermore, Grochla (1969, pp. 141f) describes 17 structuring characteristics of cooperation, but he is still not able to provide a congruent framework for the classification of cooperation forms (see also Benisch, 1972, pp. 164f).16 Nevertheless, this difficulty of rating the nature of a relationship by enumerating existing forms of inter-organizational cooperation and deriving from them structuring criteria leads to the second method of classifying cooperation forms mentioned above. The following sections present two theoretical approaches that lead to a systematization (see Section 2.1.3.3) first described by Rotering (1993, pp. 63-65).

2.1.3.1 Redistributive and Reciprocal Inter-Organizational Cooperation Several fields of research recognize a division of human interaction into “pooling” and “exchanging”.17 This division is the foundation of the approach to classify basic forms of interorganizational cooperation into redistributive and reciprocal cooperation (Rotering, 1993, p. 53). A redistributive inter-organizational cooperation exists when the participating companies seek one or several goals by pooling resources and afterwards sharing the benefits of the common activities (Tröndle, 1987, p. 19; see Figure 2). In this case, the classical distribution problem

16

For further structuring methodologies refer, for example, to Boettcher, 1974, pp. 25f; Bleicher, 1989, p. 4; Schaude, 1991, pp. 6f.

17

For the Sociology, for example, refer to Parsons, 2001, p. 72; for Jurisprudence, for example, refer to Hueck, 2003, p. 27; for the general differentiation between “pooling” and “exchanging” in other fields of research refer to Hansen et al., 1983, p. 6.

18

2. Inter-Organizational Cooperation and Supply Chain Management

arises: Participating companies have to agree on how to distribute either the commonly achieved profits or the originated loss (Eschenburg, 1971, pp. 71f; Hansen et al., 1983, pp. 6f).

organization A

organization B

redistributive cooperation

Share of A

common profit or loss

Share of B

Figure 2: Redistributive Cooperation (Source: Based on Tröndle, 1987, p. 19)

Redistributive cooperation requires of the collaborating firms to settle clearly defined agreements on both the cooperation input (what each partner will contribute) and the cooperation output (what each partner will receive from the common earnings). This class of cooperation can bring about problems if the benefits of the collaboration cannot easily be split as it is the case if knowledge is the gain of cooperation. On the other hand, and assuming that the distribution problem can be solved at all, cooperation with easy-to-split earnings (such as capacity utilization) might be less stable than cooperation with hard-to-split payback since the probability that a partner quits the collaboration is higher in the first case (Kogut, 1989, p. 188). However, reciprocal cooperation demands the exchange of goods (in terms of something valuable for the collaborating partner (see Figure 3). The cooperating organizations may, thus, be driven by different goals (Tröndle, 1987, p. 21). The partners expect to have a higher benefit of the good received from the collaborator than the loss from the good given away which requires partners to have different subjective perceptions about the value of the good to exchange.18 Here, Sahlins (1973, pp. 147f) speaks of a “balanced reciprocity.”19 In contrast to redistributive cooperation, no serious distribution problems arise in this case since the benefits of the exchange are earned separately by each organization.

18

Rotering (1993, p. 56) gives an example for a reciprocal cooperation. He describes how Lufthansa AG and Air France performed an exchange of higher management personnel for the period of some months.

19

Other reciprocities described by Sahlins (1973, pp. 147f) are “generalized reciprocity” which are transactions that are altruistic, and “negative reciprocity” which is “the attempt to get something for nothing with impunity.”

2.1 Inter-Organizational Cooperation

19

good exchanged by A

organization A

organization B

good exchanged by B

Figure 3: Reciprocal Cooperation (Source: Based on Tröndle, 1987, p. 21)

Focusing on the transaction costs, the redistributive cooperation requires an agreement on both cooperation input and output while the reciprocal cooperation only demands a settlement on the cooperation input. Thus, it can be assumed that the involved companies in the first case incur higher bargaining and agreement costs than in the second case (Rotering, 1993, p. 56). On the other hand, in the redistributive cooperation partners seek a common goal expressed by the common earnings. This can allow one to assume that the coordination and monitoring costs in a redistributive cooperation might be lower than in a reciprocal cooperation where the collaborating organizations “just” expect the partners to fulfill their obligations incurred by the contract (Rotering, 1993, p. 56).

2.1.3.2 X and Y Inter-Organizational Cooperation A second approach for classifying inter-organizational cooperation is based on the types of activities involved in the collaboration (Hauser, 1981, pp. 180f; Pucik, 1988, pp. 79f; Porter & Fuller, 1986, pp. 334f). Porter and Fuller (1986, pp. 334f) affirm that every cooperation which carries out value-adding activities can be assigned to one of the following classes: •

X Cooperation. Cooperation covers different activities within the respective cooperation partners. 3M (manufacturer) and Schneider National (logistics service provider) are an example for this cooperation type (Bowersox, 1990, p. 36). They cooperated by combining the technical leadership of the carrier with the external logistics of the manufacturer (see Figure 5).

•

Y Cooperation. Cooperation covers the same activity (or activities) within the respective cooperation partners. Examples for this kind of cooperation are purchasing pools and electronic purchasing platforms as well as the joint operation of warehouses for distribution purposes (see Figure 4).

This differentiation is based on the Value Chain model (Porter, 1986, pp. 59f, 95-97, 180-182, 307f; Porter, 1986a, pp. 13f). It describes how all company functions can be categorized into nine

20

2. Inter-Organizational Cooperation and Supply Chain Management

basic activities; no matter which sector or industry the company belongs to. These nine basic activities are divided into primary activities – such as those directly involved in the production of the good or the creation of the service, the delivery to the customer, marketing activities, and after-sales services – and support activities that are provided with required input factors and grant the infrastructure for performing primary activities. Figure 4 shows a Y Cooperation in the context of Porter’s Value Chain. Company A Support Activities

Infrastructure Human Resource Management Technology Development

Inbound Outbound Marketing Operations Logistics Logistics and Sales

Company B Support Activities

Margin Service cooperation

Primary Activities

Procurement

Infrastructure Human Resource Management Technology Development

Primary Activities

Procurement Inbound Outbound Marketing Operations Logistics Logistics and Sales

Margin Service

Figure 4: Y Cooperation (Source: Based on Porter, 1989, p. 23)

Figure 4 shows companies A and B collaborating within the Purchasing activity, e.g. in the form of joint purchasing in order to have access to higher discounts than with the lower separated purchasing volume. Figure 5, on the other hand, shows the X Cooperation between 3M and Schneider National. The latter provides technological leadership through EDI (Electronic Data Interchange) and the first puts both internal and external logistics into the hands of Schneider National. It can be assumed generally that inter-organizational cooperation of type X is more likely to take place between organizations of different strategic sectors, such as manufacturers and logistics service providers, which complement each other due to their different characteristics and skills. Also, it is plausible to assume that Y Cooperation is more likely to be found between companies within the same strategic group (e.g. suppliers form a purchasing pool, distributors share transportation capacities, OEMs create an e-purchasing platform, etc.).

2.1 Inter-Organizational Cooperation

21

Schneider National Support Activities

Infrastructure Human Resource Management Technology Development Margin

Inbound Outbound Marketing Operations Logistics Logistics and Sales

Service

on erati

3M

coo pera tion

coop

Primary Activities

Procurement

Support Activities

Infrastructure Human Resource Management Technology Development

Primary Activities

Procurement

Margin

Inbound Outbound Marketing Operations Logistics Logistics and Sales

Service

Figure 5: X Cooperation (Source: Based on Porter, 1989, p. 23)

2.1.3.3 Rotering Matrix for Identifying the Type of Inter-Organizational Cooperation Rotering (1993, pp. 63f) combines the classifications presented in the preceding sections and derives four cooperation types as shown in Figure 6: Y Cooperation

X Cooperation

Redistributive Cooperation

Cooperation Type I

Cooperation Type II

Reciprocal Cooperation

Cooperation Type III

Cooperation Type IV

Figure 6: Rotering Matrix for Classifying Inter-Organizational Cooperation (Source: Based on Rotering, 1993, p. 63)

Cooperation Type I is, thus, a resource pooling cooperation within homogeneous activities. Here, the coordination effort is comparably low, and the partners have to agree on the output shares. This is the case within purchasing pools and transportation capacity sharing where partners have to agree on how to distribute the achieved discounts, cost reductions, and economies of scale. Cooperation Type II is a resource pooling cooperation between different activities within the cooperating partners. In this case, the challenge is not only the allocation of the benefits, but also the coordination of the collaboration across the different activities of the partners. The joint venture Autolatina between Volkswagen and Ford represents such a cooperation type (Rotering, 1993, p. 64) in which the partners pool research and development as well as production resources and have to agree on how to distribute the created production capacities.

22

2. Inter-Organizational Cooperation and Supply Chain Management

Cooperation Type III represents an exchange-based cooperation within the same activity. This sort of cooperation does not require agreement on benefit sharing and represents a comparably small coordination challenge. A cross-licensing agreement in research and development provides an example for this kind of cooperation where partners give each other time-restricted usage rights of a technology (Rotering, 1993, p. 64). Cooperation Type IV is an exchange-based cooperation between different activities within the partners. Bosch Eisenach GmbH (automotive supplier) and Paul Günther Industrielogistik GmbH (logistics service provider) are an example for this kind of cooperation. Here, the LSP takes over the external logistics of Bosch while the supplier provides technological infrastructure through the merger of information systems with the LSP (Buxmann et al., 2004, p. 163). This cooperation is reciprocal since the partners achieve different benefits and there is no common output to be shared. On the one hand, this kind of collaboration allows for an easier identification of synergies due to the different characteristics and skills of the partners. But it presents a major management challenge (Rotering, 1993, p. 64). The argumentation presented in this section serves as the theoretical framework for the discussion of cooperative planning within SCM. Most of the examples to be presented in this dissertation and the context for the model evaluations in Chapter 3 describe inter-organizational cooperation of Type I. In Type I cooperation, organizations pool resources (e.g. trucks) within a homogeneous activity (e.g. transportation/distribution/external logistics) and agree on the benefit shares (e.g. reduced transportation costs/improved transportation capacity utilization).

2.2 Supply Chain Management In this section, the multidisciplinary origin of the concept “supply chain management” will be discussed, followed by an effort to find a suitable definition for SCM (see Section 2.2.2). Afterwards in Section 2.2.3, the most important objectives of the concept are presented. In Section 2.2.4, two important issues in supply chains are described: Co-opetition and the bullwhip effect.

2.2.1 Supply Chain Management as a Field of Research and of Practical Endeavors The term “supply chain management” was originally coined in 1982 by the two consultants Oliver and Webber who pointed out that businesses could potentially derive benefits from integrating the internal business functions of purchasing, manufacturing, sales, and distribution (Oliver & Webber, 1982; Harland, 1996, p. S63; Mentzer et al., 2001, p. 1 Lambert, 2002, p. 3). Today, the term is almost omnipresent in both academia and industry.

2.2 Supply Chain Management

23

SCM has already been an en vogue topic in the 1990s and the assumption that its popularity has risen to even higher levels by now does not seem too farfetched. However, it has to be pointed out that – despite all the buzz that is being raised about SCM – there is only a “relatively poor supply of empirically validated models explaining the scope and form of SCM, its costs and benefits” (Croom et al., 2000, p. 69). It is probably not too daring to state that SCM as a field of research is still in its infancy. Not only seem the number of empirically validated SCM models to be relatively limited; but even more elementarily, there is neither a uniform definition of the term, nor a common understanding of SCM as a concept (Harland, 1996, p. S63). Moreover, a confusing plethora of labels refers to both the supply chain and its management (Croom et al., 2000, p. 68), e.g.: integrated purchasing strategy (Burt, 1984), supplier integration (Dyer et al., 1998), buyer-supplier partnership (Lamming, 1993), strategic supplier alliances (Lewis, 1995), supply chain synchronization (Tan et al., 1998), network supply chain (Nassimbeni, 1998), valueadded chain (Lee & Billington, 1992), lean chain approach (New & Ramsay, 1995), supply pipeline management (Farmer & van Amstel, 1990), supply network (Nishiguchi, 1994), and value stream management (Hines et al., 2000). Despite the extensive body of academic and popular literature on the topic, there are hardly any examples of successful SCM implementations; and as a matter of fact, in many companies SCM is either non-existing or still in the infancy stage (Fawcett & Magnan, 2002).20 In other words, there seems to be a gap between discussions led in the world of academia and world of practice: Apparently, the academic discussion surges ahead, and the practical implementation leaps behind. The fact that neither a universal definition nor a clear understanding of the concept exists could possibly be explained with the multidisciplinarity of the concept SCM. As such, it draws from many different bodies of research, “which, to date, have remained largely unconnected” (Harland, 1996, p. S63). Consequently, SCM has been viewed and considered from many different perspectives (Croom et al., 2000, p. 69; Giannoccaro & Pontrandolfo, 2001, p. 1); and consequently, it can be characterized as multi-faceted and complex. Table 1 lists selected streams of research that have contributed to the field of SCM.

20

A survey conducted 2002 in the European automotive industry by the Chair of Information Systems at the Freiberg University of Technology revealed that only 26% of the participating companies had been active in the field of SCM for more than three years (see Appendix D.6, Figure 62 and Section 3.2).

24

2. Inter-Organizational Cooperation and Supply Chain Management

Table 1: Streams of Research Contributing to the Field of Supply Chain Management Streams of research

Sample source

Industrial dynamics

Forrester, 1961; Towill, 1996

Value theory

Porter, 1985

Network literature

Thorelli, 1986; Jarillo, 1988

Market channel theory

Stern & El-Ansary, 1992

Business logistics

Slater, 1976; Bowersox & Closs, 1996; Gentry, 1996; Walton, 1996; Christopher, 1998

Strategic management

Cox, 1997; Dyer et al., 1998

Inter-organizational behavior

Harland, 1996

Operations management

Clark & Scarf, 1960; Revelle & Laporte, 1996; Fisher, 1997

Information management

Buxmann et al., 2002; Friedman, 2002; Chen et al., 2003

(Source: Based on Ganeshan et al., 1999, p. 845; Giannoccaro & Pontrandolfo, 2001, p. 1)

This dissertation is based primarily on business logistics literature because it focuses on cooperation scenarios for the logistical issues of transportation whereby logistics is understood as a business function involved in SCM (see Section 2.3).

2.2.2 Defining Supply Chain Management As noted in Section 2.2.1, a single, generally accepted definition of SCM does not exist. The abundance of definitions begins already with the term “supply chain” as shows the table below: Table 2: Sample Definitions for “Supply Chain” Author

Definition

Cavinato, 1991

“Supply chains .... are popular interfirm linkages to attain joint cost savings, product enhancements, and competitive services”

Ellram, 1991

“A network of firms interacting to deliver product or service to the end customer, linking flows from raw material supply to final delivery.”

Lee & Billington, 1992

“Networks of manufacturing and distribution sites that procure raw materials, transform them into intermediate and finished products, and distribute the finished products to customers.”

Saunders, 1995

“External Chain is the total chain of exchange from original source of raw material, through the various firms involved in extracting and processing raw materials, manufacturing, assembling, distributing and retailing to ultimate end customers.”

Kopczak, 1997

“The set of entities, including suppliers, logistics service providers, manufacturers, distributors and resellers, through which materials, products and information flow.”

Mabert & Venkataramanan, 1998

“Supply chain is the network of facilities and activities that performs the functions of product development, procurement of material from vendors, the movement of materials between facilities, the manufacturing of products, the distribution of finished goods to customers, and after-market support for sustainment.”

(Source: Based on Croom et al., 2000, p. 69; Kotzab & Otto, n.d.; Mentzer et al., 2001, p. 6)

2.2 Supply Chain Management

25

The term “Supply Chain Management” has numerous definitions differing in various aspects such as, for instance, the scope of the concept as well as the emphasis on certain involved functions and processes (see Table 3). Table 3: Sample Definitions for “Supply Chain Management” Author

Definition

Houlihan, 1985

“Supply Chain Management covers the flow of goods from supplier through manufacturing and distribution chains to the end user [...]. It views the supply chain as a single entity rather than relegating fragmented responsibility for various segments in the supply chain to functional areas [...]”.

Jones & Riley, 1985

“Supply Chain Management deals with the total flow of materials from suppliers through end-users.”

Stevens, 1989

SCM is “a connected series of activities which is concerned with planning, coordinating and controlling materials, parts, and finished goods from supplier to customer. It is concerned with two distinct flows (material and information) through the organization”.

Ellram & Cooper, SCM is “an integrative philosophy to manage the total flow of a distribution channel 1990 from supplier to the ultimate user”. Christopher, 1998a

“The management of upstream and downstream relationships with suppliers and customers to deliver superior customer value at less cost to the supply chain as a whole.”

Ellram & Cooper, “Supply chain management has been characterized as a cross between traditional, open 1993 market relationships and full vertical integration. As such, supply chain management represents an innovative way to compete in today's ever changing global economy.” La Londe & Masters, 1994

“Supply chain strategy includes: ‘two or more firms in a supply chain entering into a long-term agreement; […] the development of trust and commitment to the relationship; […] the integration of logistics activities involving the sharing of demand and sales data; […] the potential for a shift in the locus of control of the logistics process.’”

Carter & Ferrin, 1995

“Supply chain management (SCM) is an integrative approach for planning and controlling the flow of materials from suppliers to end users.”

Bowersox & Closs, 1996

“The basic notion of supply chain management is grounded on the belief that efficiency can be improved by sharing information and by joint planning [ ... ] an overall supply chain focusing on integrated management of all logistical operations from original supplier procurement to final consumer acceptance.”

Bowersox, 1997

“Supply Chain Management is a collaborative-based strategy to link cross-enterprise business operations to achieve a shared vision of market opportunity”

Cooper et al., 1997b

“The integration of all key business processes across the supply chain is what we are calling supply chain management”

Metz, 1997

“Integrated Supply Chain Management is a process-oriented, integrated approach to procuring, producing and delivering products and services to customers.”

Tan et al., 1998

“encompasses materials/supply management from the supply of basic raw materials to final product (and possible recycling or re-use).”

(Source: Based on Croom et al., 2000; Kotzab & Otto, n.d.; Mentzer et al., 2001)

In order to master this abundance of definitions, various authors try to group definitions that can be found in literature. Some authors even go so far as to devote an entire paper to reviewing, categorizing, and synthesizing existing definitions (see, for example, Mentzer et al., 2001). A

26

2. Inter-Organizational Cooperation and Supply Chain Management

widely accepted classification scheme is presented by Harland (1996), who distinguishes four main uses of the term as presented in the table below: Table 4: Four Main Uses of the Term “Supply Chain” #

Use of the term “Supply Chain”

1st

“First, the internal supply chain that integrates business functions involved in the flow of materials and information from inbound to outbound ends of the business”

2nd

“Secondly, the management of dyadic or two party relationships with immediate suppliers”

3rd

“Thirdly, the management of a chain of businesses including a supplier, a supplier’s suppliers, a customer and a customer’s customer, and so on“

4th

“Fourthly, the management of a network of interconnected businesses involved in the ultimate provision of product and service packages required by end customers”

(Source: Based on Harland, 1996, p. S64)

According to Martin (2002), this classification reflects the majority of SCM literature. Looking at it, one can assume that there seems to be a general consensus that SCM deals with managing, i.e. planning, organizing, leading, and controlling, the supply chain (Carrell et al., 1996). Apparently, the core issue is the complexity of the supply chain. In other words, the point of controversy is the question which parties belong to the supply chain that is to be managed. As implied by Harland (1996) and also described by Mentzer et al. (2001), there are three levels of complexity for an external, that is an inter-firm, supply chain21 (see also Figure 7): •

Level 1 – Direct supply chain: Here, only the original equipment manufacturer (OEM), e.g. Audi, as well as its first-tier supplier and its first-tier customer are included. This corresponds with the second use of SCM as presented in Table 4. The intricacy of managing the chain is the least complex compared to Level 2 and Level 3.

•

Level 2 – Extended supply chain: This type of supply chain comprises of the OEM, firstand second-tier suppliers as well as first- and second-tier customers. This corresponds with the third use of SCM as presented above. The intricacy of managing the chain is median compared to Level 1 and Level 3.

•

Level 3 – Ultimate supply chain: Here, all suppliers starting with the ultimate supplier and all customers including the ultimate customer are part of the supply chain in addition to the OEM. Moreover, other parties, such as financial providers, third party logistics providers and market research firms, can be members. This corresponds with the fourth use of SCM

21

The first type of supply chain that Harland makes reference to is confined to the boundaries of the firm and can, thus, be described as an internal or intra-firm supply chain which can often be assumed to be less complex than an external one because fewer parties are involved.

2.2 Supply Chain Management

27

as presented in Table 4. The intricacy of managing the chain is maximum compared to Level 1 and Level 2.

Focal Company

Supplier

Customer

Flow of Material and/or Information

Intra-Firm Supply Chain

Direct Supply Chain

Extended Supply Chain Financial Provider

3rd-Party Logistics Provider

…

… Market Research Firm

Ultimate Supply Chain Ultimate Supplier

2-Tier Supplier

1-Tier Supplier

Focal Company

1-Tier Customer

2-Tier Customer

Ultimate Customer

Figure 7: Levels of Supply Chain Complexity (Source: Adapted from Mentzer et al., 2001: pp. 3f)

Considering the different approaches for defining SCM (see Table 3), this dissertation will define SCM as follows (and, thus, in accordance with the fourth type of usage of SCM and Level 3 of the SC complexity schemata): SCM is concerned with managing both the flow of materials and information (Stevens, 1989) as well as “the relationships among channel intermediaries from the point of origin of raw materials through to the final consumer” (Johnson et al., 1999: p. 5) whereby the supply chain is comprised of all suppliers, starting with the ultimate supplier and ending with the first-tier supplier, the OEM, customers, starting with the first-tier customer and ending with the ultimate customer, as well as other parties such as financial providers, third party logistics providers, and market research firms. Depending on the size of the SC, managing the relationships between all involved parties can be enormously complex. Therefore, Lambert et al. (1998) suggest classifying the relationships according to their importance to the success of the focal company (e.g. OEMs) and the supply chain in general. According to these authors, management efforts and resources of the focal company should only be allocated to the most important relationships. However, making this differentiation requires knowing exactly who the members of the supply chain are and what they contribute. In SCM literature, meticulously mapping out a supply chain and the contributions of its members is a

28

2. Inter-Organizational Cooperation and Supply Chain Management

seldomly addressed issue: Many authors seem to implicitly assume that it is known who the members of the supply chain are, and what they contribute (Lambert, 2002, p. 4ff). Obviously, various business functions as well as intra- and inter-firm processes must be involved in managing the flow of raw and processed materials, finished goods as well as information. Lambert et al. (1998), Cooper et al. (1997b), Croxton et al. (2001), and Lambert (2002) point out business functions and processes involved in SCM: Logistics, purchasing, production, research and development (R&D), marketing, and finance. Employees working in the respective business departments contribute to realizing processes such as demand management and supplier relationship management22. Since this dissertation focuses on logistical aspects of SCM, logistics is the only business function that will be looked at in greater detail (see Section 2.2). At this point, it is interesting to note that SCM and logistics were originally used as synonyms: as already stated above, Oliver and Webber, the “creators” of the term (Oliver & Webber, 1982), used SCM to “describe a new integrated logistics management approach across different business functions” (Giannoccaro and Pontrandolfo, 2001: p. 1). It is also necessary to point out that, according to Lambert (2002), logistics and purchasing are separate business functions whereas other authors, e.g. Dornier et al. (1998), consider purchasing to be one aspect of logistics (see Table 7).

2.2.3 The Objectives of Supply Chain Management Now, that SCM has been defined and its background been outlined, the question remains as to what the objectives of SCM are. The objectives of a company or of one of its parts “designate the ends thought through the actual operating procedures of the organization [or of one of its parts] and explain what the organization [or one of its parts] is actually trying to do” (Perrow, 1961, as cited in Daft, 2001, p. 53). For SCM, many different objectives can be found in literature. In this dissertation, only those objectives are presented that seem to be the ones most frequently mentioned in the relevant body of literature. The following description of these objectives, which are itemized in Table 5, is quite extensive because when striving for a goal, it is important to have a clear understanding of what its meaning is. However, most authors fail to deliver a detailed explanation of their understanding of the items that they claim to be the objectives of SCM.

22

For more information on the business functions and on the processes, see Lambert et al. (1998), Lambert (2002), Cooper et al. (1997b), Croxton et al. (2001).

2.2 Supply Chain Management

29

Table 5: Objectives of Supply Chain Management Objective

Sources

Improving customer satisfaction and service

Carter & Ferrin, 1995, p. 189; Giunipero & Brand, 1996, p. 31; Cooper et al., 1997b, p. 3; Tan et al., 1998, pp. 3f.; Simchi-Levi et al., 2002, pp. 253-256; Fawcett & Magnan, 2001, p. 12; Mentzer et al., 2001, p. 6f.; Knolmayer et al., 2002, pp. 21-23; Tan, 2002, p. 43

Lowering costs and resources required for value creation

Giunipero & Brand, 1996, p. 31; Cooper et al., 1997b, p. 3; Tan et al., 1998, pp. 3f.; Ballou, 1999; Heizer & Render, 1999; Morton, 1999; Shim & Siegel, 1999; Waller, 1999, pp. 329-358; Brewer & Speh, 2000, pp. 78ff.; Hartley, 2000, p. 27; Simchi-Levi et al., 2002, p. 126; Toomey, 2000, pp. 125-126; Monczka et al., 2001, pp. 410-517; Stevenson, 2002, pp. 174-195

Reducing inventory levels and respective costs

Cooper & Ellram, 1993, p. 14; Carter & Ferrin, 1995, p. 189; Giunipero & Brand, 1996, p. 34; Cooper et al., 1997b, p. 3; Baganha & Cohen, 1998; Tan et al., 1998, pp. 3f.; Ho, 1999, pp. 14-19; Waller, 1999, pp. 288-328; Helms et al., 2000, p. 393; Simchi-Levi et al., 2002, pp. 43-46; Toomey, 2000; Fawcett & Magnan, 2001, p. 12; Tan, 2002, p. 43

Improving efficiency and effectiveness

Giunipero & Brand, 1996, p. 34; Markland et al., 1998, pp. 602-606; Tan et al., 1998, p. 3.; Tan, 2002, p. 43

Increasing profits and profitability

Lambert et al., 1998, p. 4; Tan et al., 1998, pp. 2f.; Tan, 2002, p. 43

Increasing competitiveness and competitive advantage

Cooper et al., 1997b, p. 3; Lambert et al., 1998, p. 4; Tan et al., 1998, pp. 3f.

Improvement of cooperation

Lee et al., 1997a; Christopher & Jüttner, 2000, pp. 8-9; Monczka et al., 2001, pp. 225-265

As presented in the above table, one objective of SCM is to improve customer satisfaction whereby Spreng et al. (1996) argue that overall customer satisfaction is influenced by a customer’s assessment of the degree to which a product’s performance is perceived to have met or exceeded his or her desires and expectations. Customer satisfaction could be improved by the application of SCM since, among other things, SCM might contribute to reducing the number of stock-outs, and it might help to minimize the time span between order placement and delivery because the value creation process is streamlined by SCM. For example, customers of a car manufacturer will most likely be unsatisfied with the brand if they get the product later than promised; and they will most likely be satisfied if they get their new car on the promised day. Customers of this car manufacturer are probably even happier if the OEM is able to deliver the car faster than competitors in terms of the time span between order placement and delivery. SCM could help to achieve this goal.

30

2. Inter-Organizational Cooperation and Supply Chain Management

Another objective of SCM is to lower costs and resources required for value creation. This objective might be achieved if, for example, participants of a supply chain share resources such as their fleet of trucks which might result in improved capacity utilization and, thus, lower unit costs for transportation leading to lower total costs. Ballou (1999, pp. 135ff) and Toomey (2000, pp. 125ff), for example, deal with the topic of reducing transportation costs. Also, the reduction of production costs (Waller, 1999, pp. 329-358; Stevenson, 2002, pp. 174-195) and of purchasing costs (Simchi-Levi et al., 2002, pp. 187-189; Monczka et al., 2001, pp. 410-517) play a key role in this objective of SCM. The third objective mentioned in the above table, reducing inventory levels and respective costs, might be attained by the reduced need for holding safety stocks when SCM is implemented. Reducing inventories might be one factor contributing to the achievement of another goal, namely increasing (organizational) efficiency and (organizational) effectiveness. According to Katz and Kahn (1978: pp. 226-233; as cited in Wholey et al., 2000: p. 3), efficiency is the ratio of outputs produced by an organization to the inputs needed for making those outputs. “Efficiency is improved either by reducing the costs while holding outputs constant, by increasing outputs with the same amount of inputs, or by simultaneously reducing costs and increasing outputs” (Wholey et al., 2000: p. 4). By conducting SCM, organizational efficiency might be increased because the members of a supply chain might be able to acquire needed inputs at a lower cost whereby “cost” can be understood as the sum total of all acquisition costs which is product price plus other costs such as those for transportation and installation, and transaction costs23 related to making that purchase. Transaction costs, for instance, will decrease if a company maintains long-term relationships with its suppliers because it is not necessary to search for a new source each time a good is needed. Hence, by cooperating, i.e. conducting SCM, companies increase their own utility level, but also make a contribution to improving, i.e. to increasing, the efficiency of the entire value creation process. Markland et al. (1998, pp. 602-606), for example, describe the reduction of lead times as an improvement of efficiency. Concerning organizational effectiveness, there is no universal agreement on what this term means (Robbins, 1990, p. 49). Cameron (1986, p. 542; adapted from Cameron, 1984, p. 276) distinguishes eight different models of organizational effectiveness, each of which features a different definition.

23

Transaction costs can be defined as “[t]he costs other than the money price that are incurred in trading goods or services” (Johnson, 2000). They are the sum of “search and information costs”, “bargaining and decision costs”, and “policing and enforcement costs” (Johnson, 2000). These kinds of costs are discussed within the framework of transaction cost economics (TCE) which was developed by Coase (1937) and Williamson (1979 and 1981).

2.2 Supply Chain Management

31

Table 6: Commonly Used Models of Organizational Effectiveness Model

Definition

When Useful

An organization is effective to the extent that…

The model is most preferred when…

Goal Model

It accomplishes its stated goals.

Goals are clear, consensual, timebound, measurable.

System Resource Model

It acquires needed resources.

A clear connection exists between inputs and performance.

Internal Processes Model

It has an absence of internal strain with smooth internal functioning.

A clear connection exists between organizational processes and performance.

Strategic Constituencies Model

All strategic constituencies are at least minimally satisfied.

Constituencies have powerful influence on the organization, and it has to respond to demands.

Competing Values Model

The emphasis on criteria in the four different quadrants meets constituency preferences.

The organization is unclear about its own criteria, or change in criteria over time are of interest.

Legitimacy Model

It survives as a result of engaging in legitimate activity.

The survival or decline and demise among organizations is of interest.

Fault-Driven Model

It has an absence of faults or traits of ineffectiveness.

Criteria of effectiveness are unclear, or strategies for improvement are needed.

High Performing Systems Model

It is judged excellent relative to other similar organizations.

Comparisons among similar organizations are desired.

(Source: Based on Cameron, 1986, p. 542)

Which model and, thus, which definition of organizational effectiveness is appropriate depends on the circumstances (c.f. the right column of Table 6). For example, if goals are clear, measurable and time bound, the goal attainment model seems appropriate to determine organizational effectiveness. Within this model, the effectiveness of an organization is appraised in terms of the degree to which it achieves its goals. This model focuses on the output an organization produces; however, under certain circumstances, it might be more suitable to judge an organization by its capability to acquire input and to transform it into output as well as its ability to maintain itself internally as a social organism and to interact with its environment (Robbins, 1990: p. 58). If this applies, the systems approach might be more suitable. Within the context of SCM, this might be an appropriate model to assess organizational effectiveness: by implementing SCM in a joint effort with other companies each participant might be able to increase its ability to source input made according to certain specifications, i.e. non-standardized products or products that are not commodities, because it can work together closely with the preceding echelon to develop such specific input. Organizational effectiveness is appraised in terms of the degree to which an organization achieves its goals (Koschnick, 1995). In other words, an organization is effective if it is able to “[produce]

32

2. Inter-Organizational Cooperation and Supply Chain Management

a decided, decisive, or desired effect” (Encyclopaedia Britannica Online, 2003). For example, a company has raised its effectiveness in achieving the objective of increased customer satisfaction if it is able to reduce the number of stock-outs through the implementation of SCM practices. Improving efficiency and/or effectiveness as well as attaining the other objectives described so far might be helpful in achieving the objective of increasing profits and profitability. Profit is defined as “the surplus of revenues over costs” (Grant, 1998, p. 34) and profitability is “the quality or state of being profitable [and] […] the capacity to make a profit” (Oxford English Dictionary, n.d.). That means, a company can increase its profitability and its profit either by reducing costs (e.g. by lowering inventory levels through SCM), or by increasing revenues (e.g. by increasing customer satisfaction through SCM, assuming that a satisfied customer generates more revenue). All of the objectives described so far might contribute to achieving the last objective mentioned in Table 5: increasing competitiveness and competitive advantage. Competitiveness “usually refers to characteristics that permit a firm to compete effectively with other firms due to low cost or superior technology, perhaps internationally” (Deardorff, 2001) and competitive advantage can be defined as follows: “When two or more firms compete within the same market, one firm possesses a competitive advantage over its rivals when it earns a persistently higher rate of profit (or has the potential to earn a persistently higher rate of profit)” (Grant, 1998, p. 174). A company might attain this objective if it is able to achieve at least one of the other objectives of SCM, e.g. lowering costs through decreasing inventory levels and, thus, increasing profits. The discussion in this section showed that several different objectives can potentially be achieved through the implementation of SCM practices. Additionally, it has been shown that the goals are not independent of each other, but correlated, that is, achieving one objective might help attain another.

2.2.4 Issues Related to Cooperation in the Context of Supply Chain Management This section presents two issues related to the levels of cooperation in supply chains. The first section describes the phenomenon of co-opetition as a strategic option for collaboration between companies that might be direct or indirect competitors which opens up a number of advantages but also risks that may threaten the market position of collaborating firms. Thereafter, the bullwhip effect is presented as a symptom of lack of vertical integration in the collaborating supply chain. This effect shows the “artificial” demand variability that increases through the

2.2 Supply Chain Management

33

levels of the supply chain and is mainly a result of insufficient information sharing and inadequate coordination of methods between the partners.

2.2.4.1 Co-opetition – A Concept Describing Simultaneous Cooperation and Competition Conventionally, companies were thought to have two options when conducting business: cooperation, as described above, or competition with other companies in the same market. Brandenburger and Nalebuff (1996) introduced another option, the concept of “co-opetition” which combines these two options. Co-opetition refers to the paradox that sometimes companies seem to simultaneously cooperate and compete and, therefore, it appears to be suitable for describing the nature of many of today’s business relationships. The essence of this concept is stated in the following quote: “Business is cooperation when it comes to creating a pie [i.e. a market, the author] and competition when it comes to dividing it up [i.e. dividing up market shares, the author]” (Brandenburger & Nalebuff, 1996, p. 4). Cooperation is suitable if another company can be regarded as a complementor, that is “if it’s more attractive for a supplier to provide resources to you when it’s also supplying the other player than when it’s supplying you alone [or] if customers value your product more when they have the other player’s product than when they have your product alone” (Brandenburger & Nalebuff, 1996, pp. 18ff).

That same company, that is a complementor in one situation, can be a competitor in another whereby the other company is regarded to be a competitor “if it’s less attractive for a supplier to provide resources to you when it’s also supplying the other player than when it’s supplying you alone [or] if customers value your product less when they have the other player’s product than when they have your product alone” (Brandenburger & Nalebuff, 1996, pp. 18ff).

How co-opetition can be put into practice is demonstrated by DaimlerChrysler AG and Volkswagen AG (VW) who announced that they want to cooperate in developing their next generation of transporters (Hofmann, 2002). Furthermore, it was negotiated to determine if DaimlerChrysler AG is to produce that transporter for VW (Bertram & Hofmann, 2003). At the same time, DaimlerChrysler AG and VW are obviously competitors in the market for transporters. The strategic option of co-opetition is indeed bundled with a number of advantages which make such a partnership a valid market instrument. A survey of the Dekra Institution (Littig, 1999) of

34

2. Inter-Organizational Cooperation and Supply Chain Management

the German manufacturing, retail, and service industries gives an idea of the main factors of improvement when cooperating with competitors24: •

Bundling of resources,

•

development of products,

•

alliances against other competitors,

•

exchange of know-how,

•

more effective use of capacities, and

•

higher chances of surviving in the market.

On the other hand, the success factors that can cause the failure of the alliance include, among others: •

Mistrust,

•

the preponderance of competitive thinking over collaborative thinking,

•

mismatch between partners,

•

inadequate or insufficient communication, and

•

lack of a common goal.

In a five-year study of cooperative ventures between competitors from the United States and Japan, Europe and Japan, and the United States and Europe, Hamel et al. (1989) judged the success and failure of these partnerships by the shifts in competitive strength on each side. They, thus, discovered that Asian companies tend to enter alliances willing to learn from their partners while Western companies often seek to reduce investments (Hamel et al., 1989, p. 134). In their opinion, the key factor for collaboration to succeed is the distinct contribution of each partner in terms of skills such as basic research, product development skills, manufacturing capacity, or access to distribution (Hamel et al., 1989, p. 135). They point out that “The challenge is to share enough skills to create advantage vis-à-vis companies outside the alliance while preventing a wholesale transfer of core skills to the partner. This is a very thin line to walk.” (Hamel et al., 1989, p. 136)

24

This survey cannot be interpreted as representative due to the small size of the sample. Thus, the shares of each advantage inside the sample are not really worth mentioning. The author uses the survey, though, as an approach to explore what might be relevant aspects in the field of co-opetition.

2.2 Supply Chain Management

35

Thus, companies have to develop strategies for transferring the right amount of information to the partner, and avoiding informal or unintended exchange of know-how. The example of Western-Asian alliances highlights one of the biggest problems of cooperation within and between supply chains: vulnerability of skills to non-intended transfer: “Western companies face an inherent disadvantage because their skills are generally more vulnerable to transfer. The magnet that attracts so many companies to alliances with Asian competitors is their manufacturing excellence - a competence that is less transferable than most.” (Hamel et al., 1989, p. 136)

2.2.4.2 The Bullwhip Effect – A Frequent Problem in Supply Chains with Lower Degree of Cooperation One problem that partners in a supply chain frequently face is the so-called bullwhip effect. This effect describes the trend that “demand order variabilities in the supply chain [are] amplified as [one moves] up the supply chain” (Lee et al., 1997b: p. 93). Figure 8 visualizes this phenomenon.

Retailer‘s order to manufacturer 20

15

15

Order quantity

Order quantity

Consumer sales 20

10 5

10 5 0

0 Time

Time Manufacturer‘s order to supplier 20

15

15

Order quantity

Order quantity

Wholesaler‘s order to manufacturer 20

10 5 0 Time

10 5 0

Time

Figure 8: Increasing Demand Order Variability Upstream the Supply Chain (Source: Lee et al., 1997b)

When comparing the diagrams in the above figure, one can see that the oscillation of demand over time amplifies the further one moves up in the chain. This phenomenon is referred to as the bullwhip effect. An often stated example for this effect is that of the Pampers supply chain; Pampers are the disposable diapers manufactured by The Procter & Gamble Company (P&G). When this manufacturer examined the order pattern of the members of the Pampers supply chain, it noticed the following: Demand for diapers by end customers is fairly steady, i.e. demand

36

2. Inter-Organizational Cooperation and Supply Chain Management

variability is little. However, the distributors’ orders placed to the factory fluctuate more than retail sales. The spike in P&G’s orders of materials from suppliers varies even more (Lee et al., 1997b). Several authors (Fransoo & Wouters, 2000; Houlihan, 1987; Taylor, 1999) try to quantify the bullwhip effect, for example, by using case studies as a method. Taylor (1999), for instance, examines a multi-stage automotive component supply chain and finds evidence of increased demand variability upstream in the supply chain. His results are the following values: variance of demand is increased 0.88 times, 1.63 times, 2.17 times, 3.64 times, 3.05 times, and 13.76 times at OEM demand, final assembly, pressing, blanking, service center, and steel mill, respectively. The magnitude of the bullwhip effect depends on the specifics of a supply chain like “the retailer's ordering pattern (i.e., synchronized versus balanced orders), the demand process (i.e., stationary versus nonstationary), the channel structure (serial network versus one supplier with multiple retailers), and the inventory policy applied by the channel members, […]” (Sahin & Robinson, 2002, p. 514).

Apart from the seriousness of the bullwhip effect in a chain, common symptoms that the chain is “infected” could be inefficiencies like “excessive inventories, poor product forecasts, insufficient or excessive capacities, poor customer service due to unavailable products or long back-logs, uncertain production planning (i.e., excessive revisions), and high costs for corrections, such as for expedited shipments and overtime” (Lee et al., 1997b, p. 93).

At the core of the “bullwhip effect” problem is the distortion of information on actual end customer demand which increases the further one moves upstream in the supply chain and which results in a heightening degree of demand uncertainty and variability upstream in the supply chain. For this distortion, Lee et al. (1997b) identified four major reasons “which in concert with the chain’s infrastructure and the order managers’ rational decision making create the bullwhip effect” (Lee et al., 1997b, p. 95): demand forecast updating, order batching, price fluctuation, and rationing and shortage gaming. Each item will be explained in one of the following sections.

2.2.4.2.1 Demand Forecast Updating One reason why information on actual demand gets more and more distorted the further upstream in the supply chain one moves, is that all members of the supply chain have to make their orders based on forecasts of the expected demand of the next member down the supply chain instead of basing it on actual demand. Relatively precise information on future demand by end customers is often only available to retailers. For example, the retailer Safeway Inc. is likely to

2.2 Supply Chain Management

37

be able to forecast end customer demand for diapers with high precision; especially since demand for this product is probably relatively stable. Ideally, the wholesaler delivering Pampers to Safeway Inc. should deduce Safeway Inc.’s demand from the end customers’ demand. However, information on end customer demand is unavailable to that wholesaler, and instead, this intermediary is forced to make its plans based on forecasts of Safeway Inc.’s demand which seems to vary from the end customers’ demand. The question is why Safeway Inc.’s demand does not mirror end customer demand, and why the wholesaler’s demand does not reflect the retailer’s demand and so on. The answer to this question of why demand information gets distorted is that all demand in this chain, except the end customers’ demand, is based on the demand forecasts of the respective customers and forecasts are usually erroneous. According to Ryan (1997), the extent to which information is being distorted depends – among other things – on the forecasting technique used. There are different forecasting methods, e.g. exponential smoothing; however, no matter which of these methods is used, information on actual demand is most likely being distorted (Lee et al., 1997b).

2.2.4.2.2 Order Batching Another reason why information on actual demand gets distorted is order batching. Most likely each unit of inventory will not be replaced immediately after it is sold, but reordering will usually be postponed until a larger order quantity has accrued and/or until a certain amount of time has gone by because for each order a certain fixed amount needs to be paid (Simchi-Levi et al., 2002, p. 104) and, therefore, replacing each unit immediately would not be economically. Multiple variables influence the frequency of orders. One of them is the pursued inventory policy. For example, if a so-called (s,S)-inventory policy is pursued, reordering is only carried out in variable intervals and only if the inventory level falls below the reorder point s; the amount ordered equals S (target inventory level called “order-up-to-inventory”) minus the residual inventory (Kluck, 1998; Simchi-Levi et al., 2002, p. 103). If a (t,S)-inventory policy is pursued, variable amounts are pursued in fixed intervals (e.g. biweekly or weekly) whereby the amount ordered is, once again, equal to the difference between the target inventory level and the inventory level at the point of reordering (Kluck, 1998). A third policy is the (s,Q)-inventory policy where fixed amounts are ordered in variable time intervals (Kluck, 1998). A second variable influencing the order frequency and, thus, leading to order batching is the effort to optimize transportation in order to minimize the costs associated with each order (Simchi-Levi et al., 2002, p. 104). As show later in Section 3.1.1, it is advantageous in regard to transportation costs to transport goods in truckloads and, therefore, it seems economical to postpone orders until the order amounts to a truckload (Lee et al., 1997b).

38

2. Inter-Organizational Cooperation and Supply Chain Management

A third, closely related variable are production and transportation lot sizes. Lot sizes have the effect that only multiples of a certain amount are ordered; even though this might exceed actual demand at that time (Roland Scheidler at GEKO information meeting, June 04, 2003). For example, if a company needs 50 water pumps, but if the lot size is 150 (e.g. because 150 units fit into one container), the company is likely to order a batch of 150 water pumps instead of the needed 50 units. Last but not least, a fourth variable influencing the order frequency and, thus, resulting in order batching are “the quarterly or yearly sales quotas or incentives observed in many businesses can also result in unusually large orders observed on a periodic basis” (Simchi-Levi et al., 2002, p. 104).

2.2.4.2.3 Price Fluctuation Price fluctuations also cause the distortion of information on actual demand. If prices fluctuate, buyers are likely to attempt to stock up when prices are low (Simchi-Levi et al., 2002, p. 104) – thus, they conduct forward buying – and live on these stocks when prices are high or until they are depleted. These price fluctuations are often the result of special promotional measures by manufacturers and distributors, for example, periodical price or quantity discounts, coupons, rebates, etc. (Lee et al., 1997b). At a first glance, this seems advantageous to the buyer and, thus, possibly also to all subsequent echelons of the supply chain. When taking a closer look, however, it becomes clear that “[such] promotions can be costly to the supply chain (Buzzel et al., 1990). What happens if forward buying becomes the norm? When a product’s price is low (through direct discount or promotional schemes), a customer buys in bigger quantities than needed. When the product’s price returns to normal, the customer stops buying until it has depleted its inventory. As a result, the customer’s buying pattern does not reflect its consumption pattern, and the variation of the buying quantities is much bigger than the variation of the consumption rate – the bullwhip effect” (Lee et al., 1997b, p. 97).

2.2.4.2.4 Rationing and Shortage Gaming The fourth cause of distorted demand information is rationing and shortage gaming which occurs in situations where product demand exceeds product supply. In such situations, manufacturers might ration their products to customers. For example, if a manufacturer can satisfy only fifty percent of total demand, it might determine that each customer will only receive fifty percent of the amount ordered (Lee et al., 1997b). “Knowing that the manufacturer will ration when the product is in short supply, customers exaggerate their real needs when they order. Later, when demand cools, orders will suddenly disappear and cancellations pour in” (Lee et al., 1997b, p. 98).

2.2 Supply Chain Management

39

What appears to be an overreaction of customers to anticipated shortages is really a sound and rational economic decision: Customers “game” the potential rationing (Lee et al., 1997b, as cited in Lee et al., 1997a). “The effect of ‘gaming’ is that customers’ orders give the supplier little information on the product’s real demand, a particularly vexing problem for manufacturers in a product’s early states” (Lee et al., 1997b). The practice of gaming is not uncommon. It was, for example, exercised in the 1980s when there was a shortage of Dynamic Random Access Memory (DRAM) chips in the computer industry. There was a sharp increase in orders; however, this increase was not the result of a rise in consumption, but was caused by buyers anticipating that they might not get the total amount ordered because of rationing. Customers ordered from multiple suppliers and bought only from the supplier that delivered first; the other duplicate orders were simply canceled (Lee et al., 1997b). Now that the causes of the bullwhip effect have been described, it will be shown in Chapter 5, how to quantify this effect with the model of a simplistic supply chain, i.e. a two-stage supply chain consisting only of a retailer, that observes end customer demand, and a manufacturer, who receives orders from the retailer (Simchi-Levi et al., 2002, pp. 104ff).

2.3 Logistics Planning as Object of Inter-Organizational Cooperation As described in Section 2.2.2, various business functions are involved in managing supply chain processes. Since this dissertation focuses on logistical aspects of SCM, this is the only business function that will be looked at in greater detail.25 For this purpose, the boundaries of the term “logistics planning” are to be defined, and its characteristics and decision fields are to be described in this section.

2.3.1 Business Logistics – a Supply Chain Management Process Business logistics “describes the entire process of materials and products moving into, through, and out of a firm” (Johnson et al., 1999, p. 5). Subcategories of “business logistics” are “inbound logistics”, “material management”, and “physical distribution”26 whereby these terms can be defined as follows: “Inbound logistics covers the movement of materials received from suppliers. Material management describes the movements of materials and components within a firm. Physical

25

For a more detailed information on the business functions and on the processes, see Lambert et al. (1998), Lambert (2002), Croxton (2001).

26

Another term used synonymously is “outbound logistics”.

40

2. Inter-Organizational Cooperation and Supply Chain Management distribution refers to the movement of goods outward from the end of the assembly line to the customer.” (Johnson et al., 1999, p. 5).

Hence, one can state that a logistical process starts when a company buys input goods and ends when it sells its output. Table 7: Decision Categories in Logistics Issues

Decisions

Facilities Network

Supply chain structure Number of echelons For each echelon: Number of facilities Facility size, location, and focus Links between facilities: Information flows Sourcing patterns

Operations Process Technology

Equipment Extent of automation Investment timing

Logistics Process Technology

Storage/transportation technology Extent of information technology

Vertical Integration

Extent of integration Direction (forward/backward) Balance of capacity

Work Force

Training/recruiting Payment system Job security

Operations Planning and Control

Centralization/decentralization Computerization decision Rules inventory coverage level Location of inventory

Distribution Planning and Control

Centralization/decentralization Distribution channel selection Inventory coverage level Location of inventories

Quality

Improvement programs Control standards Measurements

Transportation Policy

Transportation modes Logistics alliances Subcontracting

Customer Service Policy

Frequency of delivery Ordering methods Pricing/discounts

Organization

Structure Reporting Support groups Performance measures

Sourcing

Purchasing Supplier selection Offshore sourcing

(Source: Adapted from Dornier et al., 1998, p. 50)

2.3 Logistics Planning as Object of Inter-Organizational Cooperation

41

Within this process, different kinds of decisions need to be made and different kinds of issues need to be resolved. Dornier et al. (1998) distinguish between twelve decision categories whereby each category comprises of different exemplary issues and decisions (see Table 7).27 This dissertation will only address one of the twelve decision categories that bear high cooperation potential and, therefore, are suitable for achieving higher cost effects in the Automotive Industry: namely, transportation (policy) in Section 3.1. Furthermore, the planning activities in which these decisions are taken can be divided into several categories that depend on their time horizon and their location within the logistic chain (Fleischmann, 2002, p. A1-A9). The next figure shows the categorization of important activities within logistics planning.

Sales

Long-term Strategic

Distribution

• Supplier selection • Alliances • Procurement system

• Distribution • Product Spectrum • Production Location Network • Service Level • Production • Markets Systems

Medium-term tactical

Production

• Headcount • Materials Procurement (contracts)

• Production Amounts • Scheduling and Configuration • Work Schedule

• Means of Transportation • Routing • Customer assignments

• Medium-term Sales Planning

Short-term operational

Procurement

• HR allocation • Material Requirements Planning

• Load Planning • Sequencing • Monitoring

• Tour Planning • Inventory Management • Handling

• Rolling Forecasts • Short-Term Sales Planning

• Strategic Sales Planning

Figure 9: Logistics Planning Activities (Source: Based on Fleischmann, 2002, p. A1-10)

The strategic planning as shown in Figure 9 determines the spectrum of efforts and activities as well as the configuration of the required logistic system. This can comprise of decisions to makeor-buy of manufacturing processes, location of factories and warehouses, investment in production assets, layout of manufacturing and warehousing surfaces, and information and communication technology (ICT). The medium-term planning sets the general conditions and capacities for manufacturing and logistics processes (e.g. personnel availability and work schedule, rough flow of materials and schedules, potential transportation relationships and customer assignment). The short-term

27

For another systematization of logistic decisions that need to be made in SCM, see Figure 1 in Appendix A.1.

42

2. Inter-Organizational Cooperation and Supply Chain Management

planning has the objective of creating instructions for the direct execution of the exact times and amounts within concrete processes. These planning activities are interrelated and, thus, each activity influences the decisions of other planning activities. For example, there is a vertical influence in the matrix shown in Figure 9 since long-term planning activities create the framework for decisions in the short-term planning activities. In addition, a horizontal influence goes from the sales to the procurement level of the logistic chain (Fleischmann, 2002, p. A1-10). In this case, there is an information flow from the market-nearer to the market-farther areas (e.g. the results of the short-term sales planning sets the framework of the tour planning for the distribution; see Figure 9). There are further influences that, for example, go in opposite directions (e.g. the sales planning has to take the planned production capacities into account, or long-term planning bases decisions on detailed costing information from operative processes). This means that, in order to reach satisfactory results, different planning activities have either to be planned simultaneously (which is probably not possible; Fleischmann, 2002, p. A1-10) or to follow a hierarchical setting with an anticipation of decisions of related planning activities. This hierarchical concept is described in the following section.

2.3.2 Logistics Planning as a Hierarchical Planning Problem Both the design and management of logistics systems and processes, which are the core tasks in logistics, can be understood as planning activities (Fleischmann, 2002, p. A1-9). As already mentioned, planning often comprises of several activities that can be dependent from each other; planning efforts have, thus, an accentuated coordination aspect (see Section 2.3.1). In this context, Schneeweiss (1994) introduces the concept of hierarchical planning which forms the basis of the process of finding decisions in logistics planning. This concept describes the planning as a sequence of planning models in which the respective higher-ranked model affects the area of decision and goals of the lower-ranked models (Schneeweiss, 1992, pp. 80f). If the concept is reduced to two decision levels or activities, the higher-ranked activity can be considered as the top-level and the lower-ranked activity as the base-level (Schneeweiss, 1999, p. 3). Each activity is characterized by its decision model which comprises an area of decision, a set of goals, and the available relevant information.

2.3 Logistics Planning as Object of Inter-Organizational Cooperation

43

Figure 10: Examples of Hierarchical Situations (Source: Schneeweiss, 1999, p. 3)

Figure 10 shows examples of hierarchical settings as they could arise in logistics planning activities. All settings in the figure present asymmetric situations regarding the decision rights (a, b, f, and h), the available information (d, e, and f), or time of decision making (a, c, and i). Top-level Planning model of the top-level Anticipated

Influence 1

Decision

Estimated planning Model of the base-level

Influence 2 Base-level Planning model of the base-level

Figure 11: Structure of a Hierarchical Planning System (Source: Based on Schneeweiss, 1999, p. 7)

The inner structure of hierarchical planning can be described within a time dependent hierarchical setting as, for example, (i) in Figure 10. For a given period t0, the top-level makes a decision which influences the area of decision, set of goals, and relevant information of the base-

44

2. Inter-Organizational Cooperation and Supply Chain Management

level (“Influence 2” in Figure 11). Thus, the given decision from the top layer has implicit consequences in the optimal decision of the bottom layer in period t1. In addition, the bottom layer has also an influence on the decisions of the top layer; depending on the way the top layer anticipates the decisions of the bottom layer. There are two basic forms of anticipation of the decision taken by the base-level (Schneeweiss, 1992, pp. 37f.; Zimmer, 2001, p. 30): •

Non-reactive anticipation: This type of anticipation merely takes the estimates of the planning model of the base-level into account to make the decision of the top-level. The possible reaction to the decision of the top-level is not considered (“Influence 1” in Figure 11).

•

Reactive anticipation: This anticipation takes into account the possible reaction of the baselevel to the planning results of the top-level. If the structure of the base-level’s planning model is completely known, one speaks of exact explicit reactive anticipation. In this case, only the difference between the given information between t0 (when the top-level makes the decision) and t1 (when the base-level reacts to the decision) creates uncertainty. If, given the knowledge of the structure of the base-level, the calculation of the anticipation function can only be approximated, one speaks of approximate explicit reactive anticipation. If only parts of the decision of the base-level can be anticipated because features of the structure are not known, one speaks of an implicit reactive anticipation which only ensures robustness of the top-level’s decision.

Of course, intermediate types and subtypes of these grades of anticipation can be found in literature (Schneeweiss, 1999, p. 11). Furthermore, the relationship between the parties involved in the hierarchical planning can be divided in those involving one decision-making unit (DMU) and those in which several DMUs interact in the planning. The latter are what is called distributed planning or distributed decision making which is the focus of the planning in the context of SCM.

2.3 Logistics Planning as Object of Inter-Organizational Cooperation

45 Constructural Hierarchies

Conflict-free Hierarchical Planning

One DMU

Organizational Hierarchies

Team Situation Enforced Team Hierarchical Planning

Several DMUs

Non-Team Situation

Antagonistic Conflict

Principal-Agent Hierarchies

Hierarchical Negotiation: Compromise

Non-antagonistic Conflict

DMU: Decision-Making Unit

Hierarchical Negotiation: Conflict Resolution

Figure 12: Hierarchies in Distributed Planning (Source: Based on Schneeweiss, 1999, p. 5)

As Figure 12 shows, the one-party situation leads to a conflict-free planning context. This is the case within constructural hierarchies which are hierarchical settings with only one decision maker. Here, all levels in the decision hierarchy have the same information status. On the other hand, the multi-party situation distinguishes the team and non-team based settings. Team-based situations with multiple decision-making units are similar to one-party situations since they are conflict-free hierarchical planning problems that only differ in communication aspects (Schneeweiss, 1999, p. 4). If one takes a look at the co-opetition aspects in this context, it can be stated that “Clearly, if the conflict free levels are following different goals their improvement is favorable for all parties, i.e. even for competitive goals the parties support each other.” (Schneeweiss, 1999, p. 4)

This is one of the key criteria which support the argumentation of the co-opetition advocates. On the other hand, the environment within a supply chain, which might be understood as a much more cooperative environment than the co-opetitive, must not be characterized as a team situation since the “common wealth” of the supply chain is not always associated with the improvement of each member. Thus, local optima would have to step aside for a global optimum. This would be a non-team based case in which the levels of the hierarchy follow different goals in an egocentric manner. The non-team based situation differentiates two cases: antagonistic and non-antagonistic hierarchical setting. The distributed hierarchical planning in a non-antagonistic situation results in “fair negotiations” without opportunistic behavior that end in either a compromise (all parties

46

2. Inter-Organizational Cooperation and Supply Chain Management

decide to meet “in the middle”) or a conflict resolution (one party cedes and accepts the position of the other party). The antagonistic case reveals a situation in which parties have an asymmetrical information setting and behave opportunistically; feinting and cheating has to be taken into account. A compromise is only reached by creating the appropriate incentive structure. These so-called principal-agent hierarchies are contemplated by the principal-agent theory.28 A special case of the non-team based situation is the enforced team hierarchical planning problem. Here, the top-level has such power that it is capable of enforcing the influence of its decisions so as to force the base-level to adopt it. The conflict is solved without offering incentives, but by “forcing contract”. In the European automotive industry, this case can often be observed in the context of EDI implementations. In this case, the OEM is able to enforce the adoption of the preferred EDI standard on its business partners (suppliers and logistics service providers). Although partners might prefer another standard (or even no implementation of EDI at all), the threat of losing the OEM as a business partner provides the reason for “forcing contract”. Thus, this situation is analogous to an organizational hierarchy in which the different DMUs have different decision rights within the planning problem.

2.3.3 Inter-Organizational Logistics Planning in Supply Chains as a Hierarchical Planning Problem Following this argument and considering the definitions of supply chain presented in Section 2.2.2, the supply chain can be understood as a multi-party planning environment in which the actors are characterized by their independency and where information, goods, and financial means are exchanged between these actors. The partners within the supply chain are, thus, the decision-making units of a multi-party hierarchical planning problem. This supply chain is a sequence of planning models in which each higher-ranked model influences the lower-ranked model. The rank is defined by the point in time when the decision of each model is made. If we reduce the context of the planning to the relationship between a supplier and a OEM (Anupindi & Bassok, 1999, pp. 222f; Zimmer, 2001, p. 32), and if we assume that the OEM takes over the top-level model, we can represent the implications of the planning models as a classical hierarchical planning problem.

28

For further information on the principal-agent theory refer, for example, to Mirrless (1974, 1976), Grossmann and Hart (1983), and Holmstrom (1979).

2.3 Logistics Planning as Object of Inter-Organizational Cooperation

47

Long-term level

Influence 3 Short-term level Top-level: OEM Planning model of the OEM Anticipated

Influence 1

Decision

Estimated planning model of the supplier

Influence 2 Base-level: Supplier Planning model of the supplier

Figure 13: Hierarchical Structure of Logistics Planning in Supply Chains (Source: Based on Zimmer, 2001, p. 32)

Figure 13 shows the hierarchical relationship between supplier and OEM in the context of planning. In a first step, the framework of the operative planning is set at the strategic level. This framework defines the area of the decision of the operative level (“Influence 3” in Figure 13). Inside of these levels, the decisions are subject to another hierarchical model. Figure 13 shows that the planning problem inside the operative level is a classical hierarchical planning problem as described in the previous section. An example of such a planning problem can be seen in the context of purchasing planning in which the OEM and supplier agree on outline contracts with the forecasted long-term demands; and in the context of short-term order batching and production planning, both parties agree on the specific to-be-delivered amounts. The outline agreement provides, thus, the demand estimates that can only be exceeded or under-run within certain limitations.

2.3.4 Inter-Organizational Planning – The Approach of Wyner and Malone In order to get a better understanding of the alternative forms of inter-organizational cooperation, the author will follow the approach of Wyner and Malone (1996). This approach distinguishes between three types of decision finding based on how information is exchanged between the cooperating partners, and on the amount of decision-making units. The next figure visualizes the three forms of cooperative planning that are presented by Wyner and Malone (1996).

48

2. Inter-Organizational Cooperation and Supply Chain Management

Supply Chain Member

No Cooperation

Central Planning Authority

Decentral Cooperation

Relationship

Central Cooperation

Figure 14: Forms of Inter-Organizational Planning (Source: Based on Wyner & Malone, 1996)

In the case of “no cooperation“, each company separately conducts its own planning activities and neither party harmonizes its plans with the other members of the supply chain (Wyner & Malone, 1996). This case represents a non-cooperative environment where the hierarchical planning is merely an internal problem of each supply chain actor. The external influence between the levels does not affect the structure of other planning problems since there is no distributed decision in a common planning problem, but rather a simultaneous decision in isolated planning problems. There are several decision-making units that will not change the structure of their planning problems (e.g. their preferences) to achieve a compromise. In the context of the supply chain, this could be understood as a non-team situation with an antagonistic (or also a non-antagonistic) conflict where the compromise accepts the noncooperative planning result (see Figure 12). The second case, “decentralized cooperation”, is characterized by the exchange of information (e.g. on demand or inventory levels) between the participating parties. If necessary, members of the network adjust their plans based on the information they received from others (Wyner & Malone, 1996). In an analogy to the theory of hierarchical planning, this situation reflects the non-team situation where there is a non-antagonistic setting and partners influence other decisions through the exchange of relevant information. The base-levels in this process adapt the structure of the planning problem with the information received from the top-levels which had made a decision on the period of time earlier. And finally, but importantly, there is the so-called “central cooperation” where a central organization (e.g. the OEM) drafts a joint plan for all partners. This situation corresponds to a conflict-free team situation (or also enforced team situation) in the theory of hierarchical

2.3 Logistics Planning as Object of Inter-Organizational Cooperation

49

planning in which either an organizational or a constructural hierarchy is implied. Two settings are possible in this case (see also Figure 12): •

There are several decision-making units. The central organization anticipates the decision of all parties involved and then distributes the instructions to all base-levels who then adopt the given solution or behave as anticipated by the central organization. Towards this end, the central organization has to possess the power to establish a constructural relationship with the partners. It is also possible that contractual agreements have reached an organization-like structure between all involved parties.

•

There is a single decision-making unit. The central organization makes a decision based on an integrated planning problem with a simultaneous planning of all involved parties. This case does not require an anticipation of the decisions made by other parties. There is no distributed hierarchical planning problem; the hierarchical situation is an internal planning problem of the central organization.

In the opinion of the author, the approach of Wyner and Malone (1996) is a valid method of operationalizing the categories of inter-organizational hierarchical logistics planning. The central and decentral division corresponds to the division in one or several decision-making units of the hierarchical planning theory. Furthermore, the exchange of information in this approach corresponds to the influence of the top-level to the base-level in the hierarchical approach. In the context of this dissertation, these three settings, no cooperation, decentral cooperation, and central cooperation, will serve as the basis for the evaluation of cooperation in supply chains.

3 Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

This dissertation focuses on cooperative logistics planning within supply chains. Cooperation within other decision categories such as the standardization of the exchange of information (e.g. through agreements on common standards; Buxmann, 1996; Domschke et al., 2002; Kimms, 2003; Domschke & Wagner, 2005; Wüstner, 2005), short-term agreements in the form of transactional cooperation (that is dominated by a demand/offer or buyer/seller relationship), or assessment of outsourcing decisions (Coase, 1937; Williamson, 1979; Williamson, 1981) are not subject of discussion here. It is the realization of common logistics planning activities and also the decentralized planning with shared information that is to be the core aspect of this chapter. After having dealt with preliminary issues such as the definition of key terms in Chapter 2, the focus of this chapter will first be on the description of cooperation scenarios for transportation (Section 3.1) in supply chains. These scenarios describe how members of a supply chain could work together and, thus, possibly reap benefits such as lower costs, an increase in market share, or an increase in profits. Scenarios presented in this chapter are either described in SCM literature or were developed by the author. Second, Section 3.2 will present the results of an empirical study of the European automotive industry whose focus was to determine the level of SCM software use in this sector as a plausible indicator of cooperative behavior in supply chains.

3.1 Cooperative Transportation in Supply Chains 3.1.1 An Exemplary Decision Category in Logistics: Transportation Setting the transportation policy is one of the most important tasks in logistics since, first of all, transportation represents one of the largest controllable costs within a supply chain (Bergmann &

52

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

Rawlings, 1998, p. 369; Quinn, 2000, p. 75). Second, this policy has great impact on other aspects of the logistics system (and vice versa) and, thus, on the system’s overall success as the following examples by Schneider (1985) and Carter and Ferrin (1995) show: •

The cost of transportation is one criterion influencing the decision on where to locate plants and warehouses as well as which suppliers to select and which customer markets to serve.

•

The mode of transportation influences packaging requirements and, thus, packaging costs.

•

The type of selected carrier often determines which material handling equipment needs to be installed.

•

Shipment sizes and, thus, transportation costs are influenced by the ordering methodology: If the pursued policy seeks maximum consolidation of shipments, shipment sizes tend to be larger and, therefore, the company can most likely take advantage of volume discounts if transportation services are purchased from a carrier (Dornier et al., 1998, p. 173). The drawback of aiming at maximum consolidation of shipments is that inventory levels tend to be rather high – especially shortly after a delivery – and, thus, also inventory carrying costs. If instead there are frequent, smaller shipments, inventory levels and, thus, inventorycarrying costs might be lower, but transportation costs might increase. One can say that there is a trade-off between transportation costs and inventory carrying costs. These two types of costs need to be weighed against one another in order to determine the proper solution to the trade-off between transportation costs and inventory carrying costs (Schneider, 1985, p. 118; Carter & Ferrin, 1995, pp. 191f).

•

The mode of transportation is one factor determining inventory requirements and, thus, inventory costs; for instance, if high speed (and, thus, usually high priced) transportation is used, the level of inventory that needs to be at hand is lower: Premium transportation might allow managers to reduce safety stocks29 and could consequently help lower overall costs and improve efficiency. Additionally, inventory-in-transit and their respective costs could be lowered by using premium transportation that is fast and very reliable.

When determining the transportation policy, decisions have to be made as to the modes of transportation, the shipment sizes as well as the routing and scheduling (Dornier et al., 1998, p. 52). Five types or modes of transportation can be distinguished; namely truck, rail, air, water, and

29

Safety stocks are stocks “which protect against [unexpected] variations in demand and lead or resupply times” (Schneider, 1985, p. 118).

3.1 Cooperative Transportation in Supply Chains

53

pipeline (Johnson et al., 1999, p. 173). If two or more of these modes are combined, this is referred to as intermodal transportation (Johnson et al., 1999, p. 173). Each mode has its distinctive characteristics. The majority of manufactured products is hauled by truck or rail transport. For example, when Audi AG ships its cars to dealers or importers, 39 percent are hauled by truck directly to the dealers and 61 percent are transported by rail either to a certain train station or to a port (Emden or Bremerhaven). From the train station, cars are transported by truck to the dealership (Friedrich Bernecker, personal communication, June 26, 2003). The two modes, truck and rail, are quite similar in many respects. However, rail is slightly less costintensive and offers a slightly lower level of performance compared to shipping by truck. Therefore, lower value-weight and lower value-per-cube products, e.g. chemicals, plastics, and steel products, are transported by rail rather than truck (Ballou, 1999, pp. 142f and p. 146). The third mode of transportation, air, comes at a premium cost, but in return offers higher performance (Ballou, 1999, p. 146). Products which are transported by this mode have either high value compared to their weight or bulk, or a condition where speed of delivery is important in their distribution (Ballou, 1999, p. 156). Examples for these kinds of products are electric and electronic equipment and parts, optical instruments, and cut flowers. Products with low value per unit of weight and nonperishable goods, e.g. coal, iron, ore, or grain, are transported on waterways, the fourth mode of transportation. The fifth mode, pipelines, is used for moving gases and liquids over long distances. These last two modes of transportation are characterized by relatively low costs (Ballou, 1999, p. 146). Various factors can be decisive in determining which of these five transportation modes is preferable. According to Ballou (1999, pp. 137f and p. 181), multiple surveys have shown that the factors “costs of service”, “average transit time (speed)”, “transit time variability (dependability)”, and “loss and damages” are the most relevant to practitioners and, thus, frequently serve as the basis for modal choice. One decision which is closely connected to the modal choice is the “make-or-buy decision”. Here, the question needs to be answered if the company should maintain its own fleet of trucks, airplanes, etc., or if it should “buy” transportation services from a carrier.30 Both alternatives have

30

The “make-or-buy decision” is often debated within the framework of transaction cost economics which was first discussed in Coase’s (1937) seminal work on transaction cost and which was further developed by Williamson (1979 and 1981). According to TCE, transaction cost varies depending on the characteristics of the transaction associated with the exchange relationship, for example, asset specificity, uncertainty, and frequency. TCE argues that buyers should internalize (or make) their supply requirements if transaction costs are high which is, for example, the case if specific assets are involved in the exchange relationships because under such

54

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

advantages and disadvantages. “Make” is chosen, if the “user hopes to gain better operating performance, greater availability and capacity of transport service, and a lower cost” (Ballou, 1999, p. 149). The drawback of the “make” decision is that it requires a financial commitment: It necessitates the investment in transportation capability (e.g. a fleet of trucks if this is the chosen mode of transportation). “Make” is advantageous or necessary mostly in two situations: If the shipping volume is high, or if special service requirements cannot be adequately met by common carriers (Ballou, 1999, p. 150). In many situations, “buy” is most likely the better choice because common carriers will be able to transport goods at a lower cost since they are in a better position to generate economies of scale. For example, if freight is to be delivered to destinations located closely to each other, common carriers can consolidate31 shipments of numerous customers and deliver them in one tour. Alternatively, customers would have to transport the products on their own and, consequently, the total number of kilometers driven in order to deliver all products would be larger, thus, also the costs for gasoline or diesel and utilizing a certain means of transportation. In addition to these two types of costs, various fixed costs (e.g. administrative costs) can be allocated over a larger volume if individual shipments are consolidated by a common carrier and, therefore, unit costs are lower. Another issue is shipment sizes. According to Johnson et al. (1999, pp. 175-180), three different size categories can be distinguished: “small-volume”, “less-than-truckload (LTL)”, and “truckload (TL) as well as carload”: •

The shipping size “small-volume” is the smallest and is for small shipments like mail and parcels. Firms specializing in the carriage of these items are often referred to as parcel carriers (Johnson et al., 1999).

•

The second smallest size, “less-than-truckload”, includes shipments that on their own do not fill an entire truck, air plane, etc. – there is space for more freight. Under such circumstances, consolidation is advantageous in order to lower the unit cost. If a company decides to “buy” transportation services, it can use the services offered by different kinds of agents that help to consolidate shipments (Johnson et al., 1999). First of all, there are freight forwarders who “are legally considered common carriers of freight and have rights and obligations as such. They do own some equipment, but this is mainly for pickup and

circumstances the hazards of opportunism by external agents or suppliers always exist. If transaction cost is low, buyers should outsource (or buy) their requirements from suppliers. 31

To consolidate means to “assemble small shipments into a single, larger shipment” (Johnson et al., 1999, p. 560).

3.1 Cooperative Transportation in Supply Chains

55

delivery operations. They purchase long distance services from air, truck, rail, and water carriers” (Ballou, 1999 p. 149). Second, there are shippers’ associations who “are cooperative organizations operating on a non-profit basis. They are designed to perform services similar to those of freight forwarders. They act as a single shipper in order to obtain volume rates. Each shipper pays a portion of the total freight bill, based on the amount to be shipped” (Ballou, 1999 p. 149). •

Third, there are transport brokers who “are agents that bring shippers and carriers together by providing timely information about rates, routes, and capabilities. They may arrange for transportation, but assume no liability for it” (Ballou, 1999 p. 149). They might consolidate shipments before handing them over to carriers, forwarders, or shippers’ associations (Johnson et al., 1999). If a company decides to “make” its own transportation services, it cannot use the services offered by these agents which means that it cannot take advantage of lower unit costs due to consolidation. One way out of this would be to postpone32 individual shipments until a larger number of shipments has accrued – which can then be consolidated and transported at a lower unit cost.

•

The largest size category is “truckload (TL) and carload”. Here, one shipment is large enough to fill an entire truck, railcar, or container (Johnson et al., 1999).

If a company decides to “make”, i.e. to produce its own transportation services, it needs to decide on vehicle routing and scheduling. In other words, the company needs to find the best path “that a vehicle should follow through a network of roads, rail lines, shipping lanes, or air navigational routes that will minimize time or distance” (Ballou, 1999 p. 191) and it needs to determine when and how frequently to deliver (or pick up).

3.1.2 Selected Cooperative Scenarios for Transportation As noted in Section 3.1.1, transportation is a highly important decision category in business logistics because it accounts for a large proportion of total logistics costs. In the following sections, scenarios will be introduced which might help members of a supply chain to reduce costs associated with transportation by cooperating with other parties and to gain other advantages.

32

“The principle of postponement proposes that the time of shipment and the location of final product processing in the distribution of a product be delayed until a customer order is received” (Zinn & Bowersox, 1988, p. 117). There are different kinds of postponements. Zinn & Bowersox (1988), for example, differentiate five types: labeling, packaging, assembly, manufacturing, and time. For more on postponement, see e.g. Zinn & Bowersox (1988) or Heskett (1977).

56

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

3.1.2.1 Cooperation Scenario I: Engaging in a Logistics Alliance Strategic and operational alliances between suppliers and manufacturers as well as between manufacturers and retailers are becoming ubiquitous in many industries (Simchi-Levi et al., 2002, p. 149). A special kind of alliance is the logistics alliance. As already touched upon in Section 3.1.1, a company must always determine whether to make something in-house or whether to buy something on the market; that is, it must make the “makeor-buy-decision” which is also referred to as the “outsourcing-decision”. One activity that can be outsourced totally or in part is the logistics function. Outsourcing this function means to hand it over to so-called third party logistics providers33 who are experts in offering logistical services, i.e. their core competency is logistics. When engaging into such a business relationship, the company seeking logistics support and the provider offering that service form a logistics alliance34 which is “a long-term partnership arrangement between a shipper and a logistics vendor for providing a wide array of logistical services including transportation, warehousing, inventory control, distribution, and other value-added activities” (Bagchi & Virum, 1996, p. 93) whereby this arrangement can be formal or informal in nature (Bagchi & Virum, 1998). The objectives (and, thus, if achieved, the benefits) of forming such an alliance are “[lowered] distribution costs and storage operating costs [and improved] quality of customer service” (Bowersox, 1990, p. 36) as well as “improved asset utilization, increased flexibility, and access to leading edge technology“ (Richardson, 1990; Trunick, 1989, as cited in Lieb, 1992, p. 29).35 Considering the approach of Rotering (see Section 2.1.3.3), this kind of cooperation can be classified as type I (or type II if both inbound and outbound logistics are subject of the cooperation). Logistics service providers offer a growing portfolio of services to their customers and many even aim at providing the opportunity of “one-stop shopping” by proffering “total logistics service packages” (Semeijn & Vallenga, 1995, p. 27). Hence, providers offering these extensive services can be regarded as something like a supermarket for logistic services: Customers can go to a single provider and acquire all desired services from that provider (Semeijn & Vallenga, 1995) – just like one can go to a normal supermarket in order to buy all needed groceries instead of going to the bakery, the butcher, the greengrocery, and so on. In order to be able to offer such a broad range of services, providers have made substantial investments to extend their range of

33

A synonym for “third party logistics provider” is “logistics service provider” (LSP).

34

A synonym for “logistics alliance” is “third party logistics”.

35

For more on motives for forming an alliance, see Frankel and Whipple (1996).

3.1 Cooperative Transportation in Supply Chains

57

services and/or they have formed alliances with other LSPs (Evangelista & Morvillo, 1999; Semeijn & Vallenga, 1995). Table 8 offers an overview of the services rendered by LSPs. Table 8: Examples of Services Offered by Logistics Service Providers Service area: Transportation -

Basic transport Expedited emergency transport Shipment consolidation/break bulk Traffic Fleet management/operation Merge-in-transit Freight bill auditing/payment Allocation-in-transit

-

Emergency deliveries Shipment racking Handling special shipments Delivery time and consistency Product unloading and shelving Floor-ready merchandise delivery Split deliveries

-

Dedicated inventory at special locations Vendor managed inventories

-

Recycling/scrap disposal Assembly Installation Product customization/configuration

-

Advance shipping notice Coordination of multiple suppliers Inventory availability Pre-filled delivery forms Joint product planning and design Coordinated production scheduling On-line product usage and technical information

Service area: Warehousing -

Storage Order picking/packing Cross-docking

Service area: Value-added services -

Breaking bulk/repackaging/product labeling Product pre-pricing and ticketing Product returns

Service area: Information services -

Tracking and tracing EDI/data translation Customs clearance Order entry/processing Product replenishment Analysis of sales data/forecasting Real-time order information

(Source: Based on Kopczak, 1997; Zinn & Parasuraman, 1997)

Cooperation between LSPs and manufacturing companies is feasible in all of the points listed above.36 The literature is replete with examples showing that such cooperation can be practiced in order to profit by lower costs, improved customer service, etc. Table 9 lists some sample cases.

36

Cooperation with LSPs is not limited to the field of transportation as might be falsely concluded because this scenario is listed under the section for cooperation scenarios in transportation. Such cooperation is feasible in all areas and, therefore, does not properly fit any category.

58

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

Table 9: Examples of Cooperation between Manufacturers and LSPs

Ia

Cooperating companies

Duration

Lever Brothers and Distribution Centers, Inc. (DCI)

Not explicitly mentioned, but Warehousing probably long-term

Category

“DCI has built, staffed, and operates a high-tech dedicated distribution warehouse for the toiletries maker [Lever Brothers] in Columbus Ohio. The companies share the benefits and risks: if warehouse utilization falls below a certain point, Lever [Brothers] helps cover the overhead; in return, DCI shares the productivity benefits when utilization approaches full-capacity economies of scale” (Bowersox, 1990, p. 36). Ib

3M and Schneider National

Not explicitly mentioned, but Transportation probably long-term

“Schneider National furnished initial computerized scheduling and electronic data interchange for 90 Minnesota Mining & Manufacturing Company shipping locations that were revamping their transportation operations in the late 1980s. The service included coordination of freight transit and associated documentation for all motor carriers 3M was using. 3M got the benefits of the latest information technology, and Schneider gained and still enjoys the position of nationwide core carriers for 3M” (Bowersox, 1990, p. 36). Ic

Sears Business Systems and Itel Distribution Systems

Not explicitly mentioned, but Value-added services; probably long-term transportation

“Sears runs a reconfiguration room in Itel’s warehouse for modifying equipment to customers’ specifications. Itel supplies full-service logistics tying into Sears’s information network. Itel assembles basic orders, positioning equipment requiring modification in the reconfiguration room. Itel then assembles the complete order and performs the tasks necessary for timely delivery to the customer” (Bowersox, 1990, p. 38). Id

–

Not explicitly mentioned, but Transportation probably long-term

“Consider the example of a workstation, which consists of two main parts: the central processing unit (CPU) and the monitor. The monitor is purchased from a supplier in the Far East and stocked on the West Coast, near the landing point, by the LSP. The workstation is built to order in the northeastern United States by the computer company. When the computer company starts to build the computer, a ship signal is sent to the LSP. The CPU and monitor are shipped from their respective origins, combined at a merge center run by the LSP, and delivered together to the customer. Under the traditional structure, both the monitor and the CPU were stocked at a centralized warehouse, to be pulled and shipped together to fill a customer order. Implementing the merge-intransit structure involves supply chain restructuring so that the two system components are no longer stocked together. Under the new merge-in-transit structure, mixing is done at one of a network of merge centers [ … ] to which the LSP has access. Thus, inventory is either positioned farther upstream or eliminated” (Kopczak, 1997, pp. 235f). Ie

–

Not explicitly mentioned, but Warehousing; transportation probably long-term

“[S]pare parts are stocked by the LSP at a warehouse located near its transportation hub. The LSP arranges for transportation (air or truck, emergency, next day or two day delivery) based on the delivery requirements as provided by the computer company” (Kopczak, 1997, p. 237). If

–

Not explicitly mentioned, but probably long-term

Transportation, value-added services, warehousing

“One partnership was based on the implementation of a supplier hub. In this case, the computer company hired a freight forwarder to design, implement and run a warehouse for components inventory to be used to manufacture the product. The freight forwarder was responsible for transportation, customs clearance, tracking/tracing of incoming freight, warehousing, and shuttling of product to the nearby manufacturing site” (Kopczak, 1997, p. 239). Ig

Whirlpool and ERX

Not explicitly mentioned, but probably long-term

Transportation, warehousing

“ERX is ‘responsible for in-bounding out product, storing it, handling it properly, and maintaining accurate inventories at all times.’ At customer sites, ERX deals with unloading, installation in some cases, and damage negotiation” (Lambert, Emmelhainz, & Gardner, 1999, p.11). That means the carrier’s driver is authorized to evaluate damages and a damage adjuster is no longer needed (Lambert et al.,1999).

3.1 Cooperative Transportation in Supply Chains Cooperating companies Ih

Xerox and Ryder System, Inc.

Duration Not explicitly mentioned, but probably long-term

59 Category Transportation, value-added services, warehousing

“Within this relationship [between Xerox and Ryder System], Ryder performs many of Xerox’s logistical tasks including, ‘supplying warehouse equipment, performing pre-installation assembly tasks, delivering and installing product, training Xerox customers on the use of the equipment, and removal of old equipment and preparation for shipping to specific Xerox locations’” (Langley & Holcomb 1992, as cited in Mitchell et al., 1992, p. 21). Ii

Coors Brewing Company and Burlington Northern Railroad

Not explicitly mentioned, but probably long-term

Transportation

“Coors created a network of satellite public warehouses and company facilities all connected to the company with high volume rail delivery. Burlington Northern Railroad served as the primary rail carrier. Under the new plan, railcars would no longer be going to small, off-line distributors. The fleet of 2,150 insulated boxcars would only be routed to 27 warehouses and about 26 distributors, all located on major rail lines. Cars would be loaded quickly, and they would be turned around quickly at their destinations. After cars are unloaded in the east, many could be sent on to Coors’ plants in Memphis and Virginia to be loaded with products headed West” (Foster, 1993, pp. 50f). Ij

Ryder Dedicated Logistics and General Motors’ Saturn division

Not explicitly mentioned, but probably long-term

Transportation

“The partnership between Ryder Dedicated Logistics and General Motors’ Saturn division is a good example of these benefits. Saturn focuses on automobile manufacturing and Ryder manages most of Saturn’s other logistics considerations. Ryder deals with vendors, delivers parts to the Saturn factory in Spring Hill, Tennessee, and delivers finished vehicles to the dealers. Saturn orders parts using electronic data interchange (EDI), sends the same information to Ryder. Ryder makes all the necessary pickups from 300 different suppliers in the United States, Canada, and Mexico, using special decision-support software to effectively plan routes to minimize transportation costs” (Davis, 1995, as cited in Simchi-Levi et al., 2002, p. 150). Ik

British Petroleum (BP), Chevron Corp., Atlas Supply and GATX

Not explicitly mentioned, but Transportation; warehousing probably long-term

“British Petroleum (BP) and Chevron Corp. also wished to stick to their core competencies. To do this, they formed Atlas Supply, a partnership of about 80 suppliers, to deliver items such as spark plugs, tires, windowwashing fluid, belts, and antifreeze to their 6,500 service stations. Rather than use the distribution networks of either BP and Chevron or create a new one, Atlas outsourced all logistics to GATX, which is responsible for running five distribution centers and maintaining inventory of 6,500 SKUs at each service station. Each service station orders supplies through its oil company, which forwards the order to Atlas and then to GATX. Each station has a preassigned ordering day to avoid system bottlenecks. GATX systems determine appropriate routes and configurations and transmit orders to the DC [distribution center, the author]. The next day, the DC selects and packs the orders, and trucks are loaded in the appropriate order based on the delivery schedule. As deliveries are made, returns and deliveries from Atlas suppliers are picked up. GATX electronically informs Atlas, Chevron, and BP of the status of all deliveries. The companies save enough on transportation costs alone to justify this partnership, and the two oil companies have managed to reduce the number of DC’s from 5 to 12 and significantly improve service levels” (Andel, 1995, as cited in Simchi-Levi et al., 2002, p. 150). Il

Simmons Company and Ryder Dedicated Logistics

Not explicitly mentioned, but Transportation; value-added probably long-term services

“Working with the Simmons Company, a mattress manufacturer, Ryder Dedicated Logistics provided new technology that allowed Simmons to completely change the way it does business. Before its involvement with Ryder, Simmons warehoused between 20,000 and 50,000 mattresses at each of its manufacturing facilities to meet customer demand in a timely fashion. Now, Ryder maintains an on-site logistics manager at Simmons’ manufacturing plant. When orders arrive, the logistics manager uses special software to design an optimal sequence and route to deliver the mattresses to customers. This logistics plan is then transmitted to the factory floor where the mattresses are manufactured in the exact quantity, style, and sequence required – all in time for the shipment. This logistics partnership has virtually eliminated the need for Simmons to hold inventory at all” (Davis, 1995, as cited in Simchi-Levi et al., 2002, p. 151).

The large number of examples shows that cooperation with LSPs is a hot topic in logistics. Various factors, such as increasingly complex supply chains, have contributed to a growing

60

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

demand for logistic services rendered by a third party. As a result, third party logistics is now a significant and still growing segment of the logistics industry (Kopczak, 1999).

3.1.2.2 Cooperation scenario II: Supply Chain-Wide Container Management Sometimes, different modes of transportation are used to deliver parts or finished goods to their final destination and/or to different parties involved in this delivery. In such situations, one problem which might arise is the necessity to repack and/or to reload the freight when switching modes or when transferring the freight from one party to another. Repacking and reloading requires material handling capacities, creates extra costs and takes time. These costs associated with repacking and reloading could be lowered by introducing a supply chain-wide container management system which would include two aspects: first, standardizing repacking requirements throughout the system; and second, letting containers circulate throughout the system instead of repacking the cargo each time as they are transferred from one party to another. Heskett demonstrates how this scenario can work if it is implemented: “Instead of merely replacing its ships with more modern versions of the same design, the company instead is converting its entire distribution system to one using containers. This system requires that orders processed in Puerto Rico be shipped in containers that will be delivered directly to customers in the eastern United States by a combination of river barge, rail, and truck. As a result (1) repackaging at all inland terminals eventually will be eliminated, (2) material handling costs and capacities at Gulf and East Coast port facilities will be greatly reduced, [ … ]” (Heskett, 1977, p. 86).

Heskett (1977) points out two possible benefits of this scenario: It is no longer necessary to provide resources for repackaging and, moreover, only one type of unloading equipment is required for the entire system. Considering the approach of Rotering (see Section 2.1.3.3), this kind of cooperation can be classified as type III (or type IV if both inbound and outbound logistics are subject of the cooperation).

3.1.2.3 Cooperation Scenario III: Selling Excess Transportation Capacity to Other Companies One drawback of in-sourcing transportation is that optimal utilization of one’s fleet cannot be guaranteed at all times. Underutilization instantly results in an undesirable rise in unit costs because overhead costs need to be covered regardless of the utilization. This problem could possibly be resolved by selling excess capacity to other companies as is, for example, done by McKee Foods Corporation: “McKee, makers of Little Debbie snack cakes, knew it could improve efficiency if it found freight to fill unused capacity on its private fleet. And the Lance snack food people sometimes

3.1 Cooperative Transportation in Supply Chains

61

find it more effective to use another fleet than their own because of efficiencies. Third party Ellcar Integrated Logistics looked into its database and saw how closely the needs of these two companies matched. So it arranged a marriage. Little Debbie trucks now haul raw materials from a number of sites in the US to Lance production facilities. Lance gets timely delivery, even when other carriers don’t have capacity available. In some lanes, sharing the same truck with a competitor has reduced transportation costs 10% to 27%” (Richardson, 1998, p. 110).

By selling transportation capacity, partial truckloads can be consolidated into a truckload and, thus, items can be transported at a lower unit cost because the total cost of transportation can be distributed over a larger number of items. The amount saved by transporting freight in truckloads can be invested, if desired, in more expensive, but faster modes of transportation, e.g. air instead of truck (see Section 3.1.1). By choosing this option, participating companies do not actually cut transportation related costs; but instead, they might be able to increase their competitiveness because today companies often do not compete solely based on price or quality but also on time37. Furthermore, by using faster modes of transportation, transit time and, thus, inventory intransit is reduced which results in a reduction of inventory carrying costs. A transit time reduction also results in shorter lead time which is the time between placing an order and having the product available. This might contribute to eliminating the bullwhip effect in a supply chain (see Section 2.2.4.2). Other advantages derived from such a cooperation are the ability to set up regular delivery schedules as well as the ability to improve customer service by faster delivery times and by an increased frequency of deliveries because the company does not have to wait until it can fill an entire truck – instead, it sells excess capacity to others (Walter, 1975). Different cooperation constellations are feasible: •

cooperation between suppliers and manufacturers (vertical cooperation): for example, a tire maker and an automobile company,

•

cooperation between suppliers (horizontal cooperation): for example, a tire maker and a light maker or two light makers, i.e. competing companies,

•

cooperation between competing manufacturers (horizontal cooperation): for example, two automobile companies,

•

cooperation between non-competing companies: for example, an automobile and a computer company.

37

This kind of competition is called time-based competition and is, for example, discussed by Hise (1995).

62

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

There are a number of service providers, like shipper’s associations or transport brokers (see Section 3.1.1) who can assist in finding a partner needing transportation capacity. It has to be emphasized, however, that such pooling of resources with other companies should only be done if transportation does not constitute a core competency. In other words, implementing this scenario should only be done if the company’s competitive advantage is not based on transportation skills (Richardson, 1998) – or the company runs the risk of loosing its competitive edge.

3.1.2.4 Cooperation Scenario IV: Joint Ownership of Transportation Capacity Joint ownership of transportation capacity, e.g. a fleet of trucks, takes the idea of selling excess capacity one step further: Whereas selling excess capacity can be a one-time occurrence, joint ownership of transportation capacities requires a long-term relationship between participating companies. Cooperation is possible between the same parties as in the case of selling excess capacities, and the pay-offs of such cooperation are the same as well. However, just like selling excess capacity, joint ownership (and joint use) should only be an option if transportation is not the basis for the company’s competitive advantage, or it runs the risk of loosing this advantage. No examples for this scenario could be found in the literature. This cooperation can be classified as type I (see Section 2.1.3.3).

3.1.2.5 Cooperation Scenario V: Multi-Stop Shipping and Sequenced Loading When multi-stop shipping is practiced, one truck (analogous for the other transportation modes) makes multiple stops at various suppliers on its way to the manufacturer and consolidates partial loads into full loads. The combined shipments are then delivered to the manufacturer as one large shipment (Anonymous, 2000). Transportation can be done by the manufacturer, by one of its suppliers, by an LSP hired by the manufacturer, or by an LSP hired by one of the suppliers. Which solution is chosen depends on the contract between the supplier and the manufacturer because this document probably specifies who is responsible for transportation. Moreover, it depends on whether the company responsible for transportation outsources this function or not. If the suppliers are responsible for transportation, and if one of them (or an LSP acting on its behalf) carries out the multiple-stop shipping, this company needs to be reimbursed by the other suppliers because it renders a service to them. Considering the approach of Rotering (see Section 2.1.3.3), this kind of cooperation can be classified as type I.

3.1 Cooperative Transportation in Supply Chains

63

Multi-stop shipping38 is, for instance, practiced by New United Motors Manufacturing Inc. (NUMMI), a joint venture between General Motors Corporation and Toyota Motor Company Ltd: “Suppliers get a firm schedule for two weeks, and ship the same amount of parts at the same time every day. Ten trucks begin a ‘milk run’ each day, picking up parts from suppliers in the Midwest and bringing them to consolidation points in Detroit and Chicago. Supplies consolidated in Detroit are then trucked to a transloading area in Chicago, where the trucks board a flat railroad car. Every day, a ‘NUMMI train’ leaves the Windy City [Chicago, the author] terminal for Fremont” (Raia, 1988, p. 71).

MG Rover Group Ltd. and its suppliers implemented a similar system. However, before the parts are delivered to MG Rover Group Ltd., they are re-sorted and consolidated in a hub or transshipment center for sequenced delivery (Christopher, 1998a, p. 189). The advantage of multi-stop shipping is that congestion at the manufacturer’s receiving dock is cleared up since only one truck arrives instead of several. Sequenced loading means that parts from several suppliers are loaded on the truck in reverse sequence to the sequence used on the assembly line. This concept differs from sequenced delivery (see Section 3.1.2.8) inasmuch as when sequenced delivery is practiced, the freight loaded on the truck is from only one supplier; whereas when sequenced loading is carried out, the freight is from multiple suppliers. Sequenced arrival of parts (regardless of whether the truck contains freight from one supplier or from several) is advantageous to the manufacturer because the manufacturer does not need to internally commission the parts, i.e. it is not necessary to put them in the right order anymore. Therefore, parts do not need to be stocked but can be assembled right away and, thus, inventory levels as well as respective costs are reduced.39 In addition to these two advantages – fewer arrivals at the manufacturer’s receiving dock and no need for internal commissioning – savings in transportation costs for the entire chain are another benefit possibly gained from the implementation of this cooperation scenario. Transportation costs are possibly lower because partial loads are consolidated into full loads and the total amount of kilometers driven by all supply chain members is lower: Instead of everyone transporting their parts individually to the manufacturer, one truck picks up and delivers all freight at once. However, the implementation of this scenario is only reasonable if the suppliers are located relatively close to each other;

38

The concept of multiple-stop shipping is sometimes also referred to as a milk run.

39

It goes without saying that sequenced loading only makes sense if different parts or different variants of the same part (e.g. seats in different colors matching the vehicle color) are loaded on one truck, train, etc. Therefore, sequenced loading is an optional element of this scenario and the following.

64

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

otherwise, transportation costs will actually increase instead of decrease. Figure 15 visualizes this cooperation scenario.

Supplier 2 10

Supplier 2

5

Supplier 1

Supplier 1 Supplier 3

Supplier 3

15

Supplier 4

Supplier 4

20 40

50

35

20

Manufacturer

Manufacturer

Total amount of kilometers driven: 50 = 20+15+5+10

Total amount of kilometers driven: 145 = 40+50+35+20

Figure 15: Multi-Stop Shipping and Sequenced Loading

Transportation services can be rendered either by an LSP acting on behalf of one of the participants, by a fleet jointly owned by several or all participants, or by a fleet owned by one participant. In either case, it is necessary to develop a fair payment system so that each party has to pay according to the amount it shipped. To make it easier, the manufacturer could pay for all transportation costs as, for example, Audi AG does (Roland Scheidler, personal communication, April 30, 2003).

3.1.2.6 Cooperation Scenario VI: Merge-in-Transit and Sequenced Loading The scenario “merge-in-transit (MIT) and sequenced loading” is similar to “multi-stop shipping and sequenced loading”. The only difference is that instead of one truck making multiple stops at several suppliers, the suppliers’ trucks deliver parts to a merge center where freight from various suppliers is consolidated to one shipment which is then delivered to the manufacturer. Parts can be put on this one truck in the sequence required for instant assembly at the manufacturer’s site (i.e. in reverse sequence) (Dawe, 1997). This cooperation can be classified as type I (see Section 2.1.3.3). How merge-in-transit works is visualized in Figure 16.

3.1 Cooperative Transportation in Supply Chains

Supplier 1

Supplier 2

LTL

LTL

Supplier 3

65

Supplier 1

Supplier 2

Supplier 3

LTL

LTL

LTL

LTL

Merge center

TL

Manufacturer

with merge-in-transit

Manufacturer

without merge-in-transit

Figure 16: Delivery to the Manufacturer with and without Merge-in-Transit

Transportation can be carried out jointly by several or all participating parties, by an LSP acting on their behalf, by one participant, or by an LSP acting on its behalf. Here again, it is necessary to find a mode according to which a company is charged for transportation services it claimed. The same applies for the merge center.40 According to Dawe (1997), the advantages of MIT are twofold: First, it is no longer necessary to hold inventory at a consolidation point or hub; and second, the receiver does not have to accommodate several shipments (one from each source) but only a consolidated one which means that the manufacturer has less receiving and unloading capacity. Several examples of the implementation of MIT can be found in the literature – however, in these exemplary efforts it is not explicitly mentioned that the trucks are sequence loaded. These sample cases make it evident that MIT can be utilized at different stages of the supply chain, i.e. between suppliers and manufacturer or between manufacturer and consumers. Table 10 lists some examples.

40

As pointed out in Section 3.1.2.1 (Table 8), organizing MIT is a service offered by LSPs.

66

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

Table 10: Examples of the Implementation of Merge-in-Transit

VI a

Cooperating companies

Duration

Dell Computer and UPS Worldwide

Not explicitly mentioned, but LSP probably long-term

Merge-in-transit by…

“Dell manufactures the processing unit but purchases monitors and keyboards from OEMs. Upon receipt of an order, UPS Worldwide merges the shipments of the processor, monitor, and keyboard from different origin points at one of their facilities in Reno, Louisville, or Austin, delivering the entire system intact.” (Dawe, 1997). VI b

Starbucks Coffee

Not explicitly mentioned, but Unknown probably long-term

“Starbucks Coffee is opening numerous retail outlets in the US. Through benchmarking McDonald's’ new store set-up process, they created a planned release system of the equipment and materials for their new stores in five phases. A system monitors the contractor's progress and determines the delivery date for each phase. Vendor orders are then timed to arrive at a consolidation hub for the assembly of staged truckloads that arrive JIT at the store site. MIT has cut the total time for new store set-up and diminished inventory costs” (Dawe, 1997). VI c

Ryder and Xerox

Not explicitly mentioned, but LSP probably long-term

“Ryder's logistics service receives vendor shipments for Xerox from all over the world and via multiple modes into their consolidation hub in Rochester, NY. Once they've received the complete materials requirements plan, they kit the materials for a single delivery to the plant to meet the hourly production schedule” (Dawe, 1997). VI d

Skyway Freight Systems and Hewlett-Packard

Not explicitly mentioned, but LSP probably long-term

“Skyway Freight Systems, an LTL trucking company, matches monitors from Japan and Korea with workstations from New Hampshire for Hewlett-Packard. Skyway tests the monitors, packs appropriate components and instructions for the peripherals, applies customer compliance labeling, and arranges for transportation to meet the customer's required delivery dates. Skyway also receives parts, assembles PCs to order specifications, and delivers complete systems for another customer” (Dawe, 1997). VI e

The Fritz Companies and Sears

Not explicitly mentioned, but LSP probably long-term

“The Fritz Companies established an MIT service for Sears to supply stores with seasonal decorations manufactured by four vendors in China. Fritz established a vendor consolidation and sorting center in Chiwan, a southern Chinese port. The Chiwan center directs vendor releases from purchase orders, loads delivery-ready store orders into containers, obtains water or air transportation, and delivers merchandise from the containers directly to the stores. Fritz's Flex information system monitors the entire process, allowing time to handle any exceptions and providing Sears with constant visibility over the entire order” (Dawe, 1997). VI f

Cisco Systems and FedEx

Not explicitly mentioned, but LSP probably long-term

“Instead [of having to build and manage a growing number of regional warehouses], the merge-in-transit program allows Cisco to match the location of the final customer with the location of the nearest FedEx depot where suppliers hip their parts of the router. The ultimate goal is to have FedEx hold all the components until the customer order is complete. Ideally, Cisco would eliminate its stocked inventory of product parts” (Schwartz, 2001, p. 29).

It might be coincidental, but in all except one example listed in Table 10, transportation and merging is carried out by an LSP (in one case it is unknown). A possible explanation is that this constellation is more transparent and straight forward compared to the joint operation of a merge center and the joint transportation by the supplier and manufacturer. An advantage of this constellation compared to a scenario in which all transportation and merge-in-transit activities are managed by one company (either a supplier or a manufacturer) is that the system is not

3.1 Cooperative Transportation in Supply Chains

67

dependent on one member and, thus, cannot be easily blackmailed or held up by this one party. If the LSP asks for unacceptable terms and conditions, it might be easier to replace the LSP rather than a supplier or a manufacturer. Thus, the LSP’s bargaining power is not as extensive which also means that suppliers and manufacturers do not have to be afraid of being blackmailed by an LSP abusing its power. This line of argumentation is based on the hold up-problem which might arise in a principal-agent relationship41.

3.1.2.7 Cooperation Scenario VII: Cross Docking and Sequenced Loading The underlying idea of this scenario is the same as in the scenarios presented in Sections 3.1.2.5 and3.1.2.6: All aim at combining the freight of different companies in order to reduce transportation and/or inventory carrying costs. This scenario only differs in that in this case cross docking is applied for combining freight instead of multi-stop shipping or MIT. Cross docking “is the process of moving product through distribution centres without storing it. In a traditional warehouse, the product moves from receiving to storage to shipping processes. With cross docking, the product moves from receiving to shipping with little or no storage of product at the warehouse” (Apte & Viswanathan, 2000, p. 292).

There seems to be only a marginal difference between cross docking and MIT. Both concepts are based on the notion of combining orders at one point of the transportation chain, and separating them at another point without any real interruption of the flow. As visualized in Figure 17, the major difference seems to be that when cross docking is applied, combining shipments are made to multiple manufacturers whereas with MIT only one manufacturer is served.

41

In a principal-agent relationship, a principal (e.g. a manufacturer) asks an agent (e.g. an LSP) to do something for it whereby this relationship is, among other things, characterized by information asymmetries: The agent always knows more than the principal, i.e. it has an information advantage over the principal (Schmidt-Mohr, 1997). One particular type of information asymmetry is called “hidden intentions” because the principal does not know the agent’s true intentions. The agent might hold up the principal ex post contracting and after the principal has become dependent on the agent, e.g. because of having made a specific investment (Klein et al., 1978). As noted by Klein et al. (1978), Goldberg (1976) was the first to discuss the “hold up-problem”. However, he did not address this issue in the context of two or more private companies engaging into a business relationship, but focused on the “hold up-problem” in contracting relative to government regulation as a means of avoiding or reducing the threat of loss of quasi rent. A hold up might, for example, manifest itself in an excessive price increase for the service rendered by the agent. Such opportunistic behavior in the form of reneging on contracts is very likely to occur “in the presence of appropriable specialized quasi rents” (Klein et al., 1978, p. 297) which are created by specialized investments. These relationships are examined within the framework of the agency theory which is divided into positive agency theory and principal agency theory. It is a core assumption of the latter theory that principal and agent are asymmetrically informed and that problems like adverse selection and moral hazard arise as a consequence. There are different categories of asymmetric information, e.g. hidden intentions, hidden actions, hidden information, or hidden characteristics. The origins of the agency theory can be traced back to Spence and Zeckhauser (1971) as well as Ross (1973), who coined the terms “agency relationship” and “agency problem”. Other groundbreaking works in this area are Mirrless (1974, 1976), Grossmann and Hart (1983), and Holmstrom (1979) (see SchmidtMohr, 1997).

68

Supplier 1

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

Supplier 2

Supplier 3

Supplier 1

Supplier 2

Merge center

Manufacturer

merge-in-transit

Supplier 3

Cross docking

Manufacturer

Manufacturer

cross docking

Figure 17: Comparison of Merge-in-Transit and Cross Docking (Source: Based on OnePL, 2002)

Although it is not shown in the above figure, parts can, once again, be sequence loaded on the truck. The advantage of cross docking (as well as of MIT) is that dwell time between shipping and receiving is minimized which means that the freight is not stored in a warehouse or distribution center resulting in a reduction of material handling costs (Schaffer, 1998) and in inventory carrying costs. More savings accrue from the opportunity to transport goods in full truck loads without encountering the problem of increased inventory carrying costs (Apte & Viswanathan, 2000). Usually, if a company receives TL deliveries, it gets more parts than it actually needs at that particular moment which means that extra parts need to be stored until they are needed resulting in high inventory carrying costs. If cross docking is implemented, the company also gets TL deliveries; but these trucks contain a mix of parts instead of many parts of the same kind. There might be just enough parts of one kind required for production at that moment and, therefore, inventory carrying costs are lower. An additional advantage of cross docking is that the delivery of parts can be speeded up (Schaffer, 1998) because parts do not sit in a warehouse or distribution center before being delivered to the final destination. It has to be mentioned, however, that cross docking (just like merge-in-transit) is not easy to implement and to manage (Simchi-Levi et al., 2002, p. 134; Stalk et al., 1992). Moreover, the implementation of cross-docking systems necessitates significant investments (Simchi-Levi et al., 2002, p. 134). Considering the approach of Rotering (see Section 2.1.3.3), this kind of cooperation can be classified as type I (or type II if both inbound and outbound logistics are subject of the

3.1 Cooperative Transportation in Supply Chains

69

cooperation). Various examples of companies applying cross docking can be found in the literature. Examples are listed in Table 11. Table 11: Examples of the Implementation of Cross Docking

VII a

Cooperating companies

Duration

Cross docking by…

Wal-Mart

Not explicitly mentioned, but probably long-term

Unknown

“Wal-Mart uses a Hub and Spoke network to distribute its products to its retail outlets. Items from vendors arrive at a distribution centre (DC) as FTL shipments. At the DC, the FTL shipments from various suppliers are broken up and consolidated again to create FTL shipments that go to the various retail outlets. The items stay in the DC for very little time and ideally move directly from the inbound dock to the outbound dock. Cross docking has helped Wal-Mart reduce its costs and has enabled the introduction of an every-day low price (EDLP) strategy. This has helped Wal-Mart improve its market share and profitability (Stalk et al., 1992)” (Apte & Viswanathan, 2000, p. 292). VII b

Wal-Mart

Not explicitly mentioned, but probably long-term

Unknown

“In this system [cross-docking], goods are continuously delivered to Wal-Mart’s warehouses, where they are selected, repacked, and then dispatched to stores, often without ever sitting in inventory. Instead of spending valuable time in the warehouse, goods just cross from one loading dock to another in 48 hours or less. Crossdocking enables Wal-Mart to achieve the economies that come with purchasing full truckloads of goods while avoiding the usual inventory and handling costs” (Stalk et al., 1992, p. 58). VII c

Columbian Logistics (functioning as Not explicitly mentioned, but an LSP) and a discount retailer probably long-term

LSP

“The wholesaler purchases products such as toilet tissue and paper towels from four different vendors, and those vendors ship to my docks from about 10 different sites in the country [ … ]. [Columbian Logistics] then delivers these products to about 200 of the discount retailer’s stores in the Midwest” (Terreri, 2001, p. 30). VII d

Columbian Logistics (functioning as Not explicitly mentioned, but an LSP) and a candy maker probably long-term

LSP

“A major candy manufacturer will ship a truckload of about 200 different SKUs to [Columbian Logistics’] temperature-controlled cross dock in Detroit [ … ]. [Columbian Logistics] then take[s] those products and make[s] up about 30 orders going to 30 different customers. So in this case, you have one location shipping into the cross dock, and product going out to several locations” (Terreri, 2001, p. 30). VII e

Columbian Logistics (functioning as Not explicitly mentioned, but an LSP), yoghurt manufacturer, and probably long-term a grocer

LSP

“The grocer wants each pallet to have five layers, with each layer a different flavor [ … ]. The yogurt manufacturer ships pallets containing the same flavor and Columbian reconfigures pallets shipments according to the grocer’s requirements” (Terreri, 2001, pp. 30f).

As can be seen in the above examples, cross-docking and transportation can either be carried out jointly by several or all participating parties, by an LSP acting on their behalf, by one participant, or by an LSP acting on its behalf. This is similar to scenarios V und VI (see Sections 3.1.2.5 and 3.1.2.6).

3.1.2.8 Cooperation Scenario VIII: Sequenced Delivery It has been noted in Section 3.1.2.5 that sequenced loading and sequenced delivery are quite similar whereby sequenced delivery is the reception of “parts from [one supplier] in the same

70

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

sequence they are used on the assembly line” (Raia, 1988, p. 72) which can be regarded as “the ultimate form of JIT” (Raia, 1988, p. 72) because there is no need to store parts before they are assembled: They arrive just when needed for production. There are numerous examples of the implementation of sequenced delivery, e.g. the delivery of car seats at Chrysler (now DaimlerChrysler AG): “Seats were the ideal spot for the automakers to move into sequenced delivery. It takes less than an hour to make a seat, 20 minutes in a pinch […]. That gives the seat supplier plenty of time to build the seat, get it on a truck, and ship it to an assembly plant that’s probably less than an hour away. Usually the seat manufacturer has about 3-4 hours’ notice of the exact sequence of cars coming out of paint. Besides the color, he knows the seat style (bench or bucket) and option package (motorized seat adjusters, etc.) – information which is transmitted electronically from the assembly plant to the seat supplier. In fact, the seat assembly line is synchronized to move at the same pace as the car assembly line. ‘If our line slows down, then the seat supplier’s line slows down also,’ says Chrysler’s purchasing director, Thomas Stallkamp. The seat supplier loads his trucks inversely; that is, the seat for the last car coming off the line is the first to be loaded onto the truck. Special racks are used to hold the seats, and in many cases offloading is done by robots” (Raia, 1988, pp. 75f).

Another example is the sequenced delivery of tires to American Honda Motor Co., Inc.: “The sequenced delivery of tires to Honda’s Marysville plant is a marvel of precision and automation. The tires arrive already mounted on wheels in stacks of five (four tires plus the spare). The stacks are picked off the truck and carried by a conveyor to the assembly line. When the tires reach the tire-mounting station, the stack is divided – two tires placed on the left side, two on the right side of the car, while the spare continues its journey to a subsequent point on the assembly line, where it is placed in the trunk […]. The tires are sequenced to match the car models on the line – two-door or four-door, Honda Accord or Civic – which means they must be balanced accordingly. And with certain aluminum wheels, the tires must be sequenced in the stack to correspond with the front-left, rear-left, front-right, and rear-right wheels. One supplier responsible for the entire show – mounting the tires on the wheels, balancing the tires, and making sure the tires are in the right sequence when they are loaded on the truck heading to the Marysville plant every hour” (Raia, 1988, p. 76).

These two examples demonstrate the high level of precision that needs to be achieved when companies try to implement sequenced delivery. Business partners must be willing to engage in a very close and long-term oriented relationship because this scenario requires flawless coordination of each party’s actions based on an excellent communication infrastructure. Considering the approach of Rotering (see Section 2.1.3.3), this kind of cooperation can be classified as type I.

3.1 Cooperative Transportation in Supply Chains

71

This concludes the presentation of cooperation scenarios discussed in the respective literature. In the following chapter, Chapter 4, it will be contemplated if these scenarios could be implemented in the supply chain of the Audi A8 V8 4.0l diesel engine, and if benefits could be reaped by this. After this literature driven exposure of cooperative scenarios, the next section offers an empirical insight into the instrument for supporting SCM activities: SCM software.

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains – An Explorative Survey on the European Automotive Industry One of the motivations of this dissertation is to explore the status quo of cooperation in supply chains in the European automotive industry through SCM software. For this purpose, the Chair of Information Systems at the Freiberg University of Technology conducted an explorative survey from February through June 2002 in which the use of such software solutions as an instrument to support cooperative planning in supply chains was addressed.

3.2.1 Information Sharing as Premise for Cooperation in Supply Chains In supply chains, companies often face the difficult task of having to coordinate their orders for parts with their suppliers according to the demand for their end product. If they are not able to properly coordinate orders for parts according to the demand for the end product, they will either end up with excessively high inventory levels of parts or end products, or insufficient supplies of the end product – both resulting in avoidable costs. In order to be able to coordinate orders for parts according to the demand for the end product, companies need information on the other companies in the supply chain. In a picture perfect world, all members of a supply chain would have complete information, for instance, on end customer demand as well as on inventory, demand, and capacity of all supply chain members. However, in the real world, the lack of sufficient and accurate information is the norm. This is partly due to the fact that the state of complete information is utopic which can partly be attributed to the limited exchange of critical data between supply chain members. Serious problems in sourcing, e.g. over-supplies or stock-outs, can be traced back to this cause (see Section 2.2.4.2). The quote below metaphorically describes the typical interaction in a supply chain: “The general picture of a traditional supply chain is a series of clearly defined interfaces between suppliers/manufacturers, manufacturers/distributors and distributors/customers with minimal information exchange across these interfaces. Orders are thrown over the wall one way and the goods thrown over in the reverse direction” (Evans et al., 1993).

72

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

In other words, information in supply chains is often only exchanged bilaterally between two adjacent monolithic supply chain members; usually delayed in time and insufficiently (see also Chapter 4).

supplier T2

supplier T1

OEM

customer T1

flow of products and materials flow of information

Figure 18: Bilateral Information Exchange in Supply Chains

This dyadic exchange of information within the supply chain as shown in Figure 18 is the root of several problems, e.g. long information lead times (Risse et al., 2002): Information lead time per echelon amounts to about one week, and it takes up to eight weeks for all members of a supply chain with six to eight echelons to adjust to a change in the production plan of the top-level member. This results in creating safety stocks at each echelon and in the bullwhip effect (Risse et al., 2002). It, thus, follows that supply chain performance could be improved by implementing systems for a supply chain-wide, real-time exchange of information on inventory, demand, capacity, etc. allowing for a higher degree of transparency throughout the entire supply chain (see Figure 19).

supplier T2

supplier T1

OEM

customer T1

flow of products and materials flow of information

Figure 19: The Supply Chain-Wide Exchange of Information

The benefits of supply chain-wide information exchange and transparency are (Stank et al., 1999; Hellingrath, 2001; Risse et al., 2002; Scheidler, 2002; Odette, 2003):42

42

For an overview of information systems valuation methods see for example Weitzel et al. (2003).

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

73

•

Improvement of production planning processes,

•

Reduction of deterministic forecasting errors of medium- to long-term demand planning methods allowing an optimization of forecast quality,

•

Anticipation of delivery bottlenecks: Identifying problems in advance allows one to take countermeasures,

•

Shortening of replacement lead time for semi-finished goods,

•

Increased flexibility regarding adjustments to changes in the planned production program,

•

Optimization of inventory levels: Savings through reduced inventory levels as well as savings through the prevention of stock-outs when inventory levels are too low,

•

Savings through avoiding scrap since suppliers get informed more timely about discontinuations and engineering changes,

•

Savings through reducing the number of special measures (e.g. special deliveries and, hence, reduction of premium freight),

•

Increased flexibility and synchronization of the supply network,

•

Reduction of non value adding costs like trouble shooting and administrative effort to manage and control material flow,

•

Reduction of order cycle time as well as of cycle time variance.

The above list shows that information exchange not only helps companies to coordinate their orders for parts with the demand for their end products, but that this is also beneficial in many other planning respects. When talking about the benefits of information sharing, one also has to consider the costs of improving transparency in a supply chain. Often companies try to realize a smooth flow of information throughout the supply chain by implementing software applications such as SCM software. Hence, one has to contrast the benefits with the costs of installing and running such a system (Anonymous, n.d.; Hellingrath, 2001). One catchword often used in the context of improving supply chain transparency is supply chain monitoring (SCMo) which is “a multi-level SCM concept supporting the fulfillment/execution process. Basically it checks permanently if the actual inventory levels in the supply network are too high or too low regarding the demand of the next days/weeks. The multi-level approach allows to take into account the cumulated inventory across the supply chain and enables a fast and transparent

74

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry demand calculation. Every company will get early and clear signals whether to speed up or to slow down production” (Odette, 2003, p. 2).

In other words, when SCMo is implemented in a supply chain, all members provide certain information, e.g. on inventory, demand, and capacity, with which one calculates if there are sufficient, too many, or too few resources available in the system in order to satisfy a certain demand. Through SCMo, the speed of information flow between companies is increased and inter-enterprise visibility as well as synchronization are improved (Odette, 2003). An example for a SCM application supporting SCMo is ICON-SCC43 which has, for instance, been implemented by Hewlett-Packard GmbH, DaimlerChrysler AG as well as BMW AG (ICON, 2003c) and which Audi is currently testing. The objective of this tool is to facilitate the sharing of information between partners in a supply chain (e.g. on inventory, demand and capacity) and to allow them to better manage the flow of products and materials based on this information. With the help of ICON-SCC, the conventional bilateral exchange of information between partners is replaced by making information on inventory, demand, and capacity available supply chain-wide in order to foster faster, clearer real-time communication between members of the supply chain (ICON, 2003b). Figure 20 is a screenshot of ICON-SCC. It shows a multi-tier supply chain with each member being represented by a square, and the relationships between those entities is indicated by lines connecting them (for this and the following, see ICON, 2003b). Each square contains information on the respective party’s input and output buffer44 (left and right vertical status bar), in-transit inventory (semi-circle attached to the square) as well as capacity utilization (horizontal bar). Based on this information, i.e. time restrictions45 and top-level member’s gross demand46 at a certain point in time, it is calculated if the amount of inventory at a specific point in the system is too high, too low, or sufficient in the current period. The metrics used in this context are upper

43

SCC is the abbreviation for “Supply Chain Collaboration” (ICON, 2003b) and ICON Gesellschaft für SupplyChain-Management mbH (ICON) is a company located in Karlsruhe, Germany, offering various Supply Chain Management solutions (ICON, 2003a).

44

All purchased parts go into the input buffer before they are needed for production. Transformed parts are stored in the output buffer after the production process is finished and before they are delivered to the customer. Parts that are not physically changed (throughput parts) are only registered as input parts.

45

For example, transit time between two points in the system and lead time.

46

Gross demand is a company’s need for certain parts in order to build a certain amount of its products. For example, if Audi wants to build five cars, its gross demand of engines is five. Currently, orders usually do not reflect gross demand, but so-called net demand. Net demand is calculated based on gross demand: It is determined by subtracting available inventory from gross demand and by considering lot size and transportation optimization as well as other optimization measures and restrictions like container management (Roland Scheidler at GEKO information meeting, June 04, 2003; Stefan Mayer, personal communication, May 21, 2003). Supposedly, the exchange of gross demand is one of the benefits of using tools like ICON-SCC.

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

75

control limit (UCL) and lower control limit (LCL). These figures signal for how many days inventory at a certain point in the supply chain will last depending on the demand of the top-level member. If the number of days ranges between LCL and UCL, target inventory levels are met. If it is below LCL, inventory is too low; and if it is above UCL, it is too high. The inventory situation at a certain point in the system is visualized by the color of the status bars which can be green, red, or yellow. Green means that the inventory level is between LCL and UCL, red means that it is below LCL, and yellow means it is above UCL. The same logic applies to the capacity situation which is represented by the horizontal bar for each member (ICON, 2003b).

Figure 20: Screenshot from ICON-SCC (Source: ICON, 2003d)

The tool allows all members of the supply chain to get gross demand for a particular day as well as information on inventory levels at that particular day, and information on capacity utilization from other members depending on gross demand of the top-level member at that particular day. Based on this information, each company can draw its own conclusions for production planning – the calculated gross demand is only extra information, and no supply chain member is obliged to plan its production according to this. In other words, ICON-SCC is a tool for supporting supply chain planning, a tool for making the supply chain more transparent (Stefan Mayer, personal communication, August 13, 2003). In so doing, it allows all supply chain members to

76

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

make better production planning decisions47 so that problems like the bullwhip effect do not occur or do not occur as severely as they would happen otherwise. It is rarely determined if the sharing of information as is facilitated, for example, by ICON-SCC can be beneficial to each supply chain member individually as well as to the chain as a whole. Albeit attitudes towards exchanging proprietary information with other supply chain members have changed, the free flow of information still does not seem to be the norm today (Stank et al., 1999). One explanation for this reluctance to share information might be that the exclusive possession of information can constitute a source of power (Berry et al., 1994) if it is misused as such. Therefore, many companies purposely distort order information in an effort to camouflage their real intentions to their competitors, their suppliers, and their customers (Mason-Jones & Towill, 1997). Another explanation might be the lack of trust in the other members of the supply chain (Martin, 2002; Paul, 2003). Actually, some precaution does not seem to be totally out of place because the information exchanged is quite sensitive and, therefore, it is of paramount importance to ensure that all data is used only for the intended purpose, i.e. for improving supply chain performance, and that it does not get into the wrong hands (Koller, 2000). Another obstacle in realizing extensive information sharing throughout the supply chain are the systems that are used to evaluate employee performance. If employees are evaluated based on company performance and not on supply chain performance, they have incentives to do whatever is good for their company; even if it is to the detriment of the entire system. In other words, if it is good for company performance not to share information with others, employees are likely not to hand any relevant information to other supply chain members because this strategy is likely to maximize their personal utility (Sahin & Robinson, 2002). Although the sharing of information seems to be a promising method for mitigating the bullwhip effect, one should not ignore that distorted information, due to the fact that each supply chain member needs to make forecasts of their customers’ demand because this information is not available, is only one of several factors contributing to the bullwhip effect (see Section 2.2.4.2). From this, it follows that if one wants to assess the impact of sharing information, it is necessary to single out that part of the bullwhip effect which can be traced back to the lack of undistorted information at the different supply chain levels. This, however, is reported to be extremely difficult (Fransoo & Wouters, 2000; Mason-Jones & Towill, 1997). Fransoo and Wouters (2000,

47

With better understanding, it is possible to create production plans that coincide with end customer demand instead of only having a vague idea about this demand and producing too much, or too little.

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

77

pp. 84f) list some reasons why it is so troublesome to identify how much each cause of the bullwhip effect contributes to the total effect: “(1) Unclear ordering policies, meaning that order batching takes place but rules for this are unclear so rearranging data to account for the effect would not be possible. (2) No data on real demand. Separating out the effect of price fluctuations and of shortage gaming would require some data on real demand compared to sales. However, such real demand data will often not be available. (3) No data on shortages. The comparison of real demand with sales would be relevant for moments when shortages occur and prices change. Such information is not always available, especially data on shortages and delivery performance are often not recorded systematically.”

Even though information sharing is not a panacea as the above discussion demonstrates, one can still find examples of companies or rather supply chains that have profited from sharing information such as demand or inventory data. Three cases are summarized in Table 12. Table 12: Examples of Extensive Information Exchange between Companies Cooperating companies

Kind of information shared

L&TT and its suppliers

Not explicitly mentioned

L&TT – a Hong Kong based multination company, manufacturing and distributing electronic components to its customers mainly located in Europe – had trouble managing its inventory efficiently and therefore they examined how they could improve inventory levels as well as respective costs. They experimented with increasing the amount of information shared with suppliers. It was found that inventory levels as well as inventory costs were the least, the more information was shared (Yu et al., 2001, pp. 118f). Dell and its key suppliers

Consumer information, demand forecasts, quality and inventory information

Dell implemented a “Web-based supply chain management program [DSI2, the author] designed to improve the ability of its component suppliers to accurately forecast their production” (Anonymous, 2000, p. 70). Dell’s key component suppliers now get customer information and thus are able to adjust their manufacturing plans according to actual end customer demand. DSI2 is an addition to a portal that Dell uses to facilitate the exchange of quality and inventory information between Dell and its suppliers (Anonymous, 2000). “Using Internet technology, Dell can swap more robust forecasting data and receive better visibility into component availability from its key 38 suppliers […]” (Anonymous, 2000, p. 70). This comes along with several benefits like “to improve operating efficiencies, to reduce logistics and inventory expenses, and form tighter relationships with suppliers […]”(Anonymous, 2000, p. 70). Dell is also considering to pass “data directly to second- and third-tier suppliers and foundries as a way of ensuring continuous supply […]” (Anonymous, 2000, p. 70). Procter & Gamble and Wal-Mart

Consumer information

“Procter & Gamble presently take the consumer information from Wal-Mart at point of sale and decide how frequently and how much stock to deliver to the stores to meet the contracted consumer service level. It is Procter & Gamble's responsibility to keep shelves full and thus maximize Wal-Mart sales. How this is achieved is entirely up to Procter & Gamble. This approach benefits both companies individually and hence their supply chain is far more flexible. Wal-Mart not only gets a far superior service from their supplier but has eliminated buffer stock holding in individual stores. Procter & Gamble now has far better control of its factories because it can see the whole picture and knows exactly what is going on in the marketplace. This is a far cry from the usual supplierretailer relationship where the retailer determines how much stock it will order from the supplier with both parties double guessing the true state of the system” (Towill, 1994, as cited in Mason-Jones & Towill, 1997).

78

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

Information sharing supported by corresponding SCM software solutions seems to be a quite promising method for improving supply chain performance, i.e. in terms of optimally adjusting inventory levels or order quantities; however, it is not so easy to put this into practice often the implementation will require major monetary and organizational efforts and will be combined with radical change in the relationships between supply chain members. In order to explore to which extent such SCM solutions have gained access to the market, a survey was conducted and is presented in the following sections.

3.2.2 Goals of the Survey The goals of this survey were, on the one hand, to gain empirical data on the diffusion of specific SCM software solutions in order to see if network effects (as described by Katz & Shapiro, 1995) might play a role in this market. On the other hand, this study aimed at identifying possible causes for the reluctance to use SCM software for collaborative planning purposes. This study does not claim to be representative for the European automotive sector (since this is a nonprobability sample; Kalton, 1983, pp. 90f.; refer also to Goodman, 1961; Watters & Biernacki, 1989), but rather to have an explorative nature and, thus, disclose relevant aspects. The complete results are presented in Ahsen et al. (2002). Given the need for information systems to support complex activities as those involved in SCM, the question is if “traditional” business software (e.g. ERP software solutions such as SAP R/3) is able to support SCM-specific processes. The need for such a software solution is mainly based on the fact that the software which has been used for logistics planning and execution, such as ERP software, provides primarily inbound oriented functionality: ERP software is neither designed to manage networks of companies nor distributed processes (Yellurkar, 2002). It primarily complements the MRP (Material Requirements Planning) planning concept. The idea of MRP divides the problem into several separate problems that are solved in isolation which results in a failure to attain the overall optimum in most cases. On the other hand, SCM systems are based on an integrated, continuous, and constraint-oriented planning idea. This principle aims to plan the logistics activities throughout all functional areas and seeks to release synergies between companies, divisions, and other operational spheres. Although SCM is a matter of integrated planning and executing logistics processes, most SCM software vendors currently provide supply chain planning (SCP) functionalities based on ERP systems to perform the execution (SCE) of the planned tasks. This usually leads to restrictions in

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

79

supporting inter-organizational processes. Figure 21 reflects the resulting supply chain architecture. Company 1

Company 2

Company 3

ERP System

ERP System

ERP System

SCM Solution

SCM Solution

SCM Solution

Integrated Supply Chain

Figure 21: Complementary Use of ERP and SCM Solutions (Source: Buxmann et al., 2004b, p. 297)

The question arises which goals are pursued in the use of SCM software. For the following analysis, the author assumes that SCM software is implemented primarily to achieve the objectives mentioned above (see Section 2.2.3). However, the implementation of SCM software often leads to organizational changes. In this context, the software can be considered a trigger for a supply chain redesign or even an occasion to replace a functional organization by a process organization. For example, decisions have to be made on the choice of warehouse locations and production centers in order to realize the best transportation and inventory strategy. Since location decisions usually encompass many facilities, multiple products, and sources to serve multiple customers, it is recommended to apply various methods to the location planning process (refer for example to Domschke, 1996) and to implement decision aids (Ballou, 1999, pp. 564570; Mathiesson-Öjmertz & Johansson, 2000).

3.2.3 Research Design It seems that, above all, case studies on the implementation of SCM software have been published. This survey, as far as we know, is the first to offer a widespread empirical study on the use of SCM software in the European automotive industry. Thus, the survey concentrates on discovering the characteristics of this market and provide possible interpretations. These interpretations are to be the subject of further representative surveys which are to be formulated as hypotheses so as to be able to confirm or reject these hypotheses. We conducted our study between February and June 2002. Information on 1,000 companies was collected through research on the Internet, company registers, embassies of European countries

80

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

as well as government offices responsible for trade and international relations. The focus on the automotive industry ensures a concentrated analysis of research questions by eliminating interindustry variations. For our study, the automotive industry was selected primarily due to its leadership role in implementing new technologies. In addition, the decreasing manufacturing penetration of car manufacturers (McIvor et al., 1998; Mathiesson-Öjmertz & Johansson, 2000) leads to corresponding requirements concerning SCM and SCM software solutions. Our sample consists of car manufacturers, suppliers, distributors, and logistic service providers from 25 European countries. As small companies are less likely to have SCM software implemented (Darrow, 2002), companies with an annual turnover of less than € 10,000,000 were excluded from our sample in order to ensure a higher response rate. Because of this selection in our survey, we do not differentiate between smaller and larger companies. First, we approached the companies either via e-mail, phone, or fax and asked them to name the responsible person for the logistics or SCM department/efforts. The companies were supplied with the hyperlink to our online questionnaire, or received the questionnaire via fax or mail. After two follow-up procedures, we had a total of 178 usable answers. This number equals a response rate of 17.8%. The survey is divided in two main areas: First, attention is focused on the cooperative behavior of companies in European industry (see Section 3.2.4). Second, the survey concentrates on the use of SCM software as a possible indicator of the application of SCM concepts in an organization (see Section 3.2.5). Here, the survey tries to provide an insight into the goals pursued by SCM software implementation (Section 3.2.5.2) and its obtained benefits (Section 3.2.5.3). Finally, Section 3.2.5.4 presents the results of the survey concerning the possible network effects in the SCM software market.

3.2.4 Cooperation in the European Automotive Industry Most of the companies in our sample are part of more than one supply chain (see Figure 22). These results indicate that co-opetition (Brandenburger & Nalebuff, 1996) appears to be a real issue, and problems such as conflicting priorities and trust which can disturb the flow of information for cooperative activities have, in fact, to be addressed. In this regard, the number of companies that chose not to answer the question, i.e. 26%, may indicate how sensitive supply chain partners deal with the topic co-opetition and its effects. Overall, the responses to this question highlight the problem of co-opetition as a palpable issue in the European automotive industry.

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

81

12% 26%

one supply chain more than one supply chain undeclared

N= 178

62%

Figure 22: Is Your Company Taking Part in Multiple Supply Chains?

3.2.4.1 Fields of Cooperation In theory, a supply chain should be an extremely cooperative environment since business partners share common goals and use similar performance measures (Premkumar, 2000, p. 59). Thus, we examined the cooperative behavior patterns of the participating companies. In this context, the term cooperation can, for example, describe either a single project or a long-term collaboration in the fields of R&D, e-commerce/e-business, and development of standards as well as manufacturing, inventory, and the consolidation of transportation. Companies in the European automotive industry seem to be remarkably cooperative since 95.5% collaborate at least on one of the mentioned levels with at least one partner in the supply chain. A high percentage of 77% cooperates with supply chain partners in the field of R&D, and about two thirds of all participating companies cooperate with partners at the levels “consolidation of inventory,” “consolidation of transportation,” and “manufacturing.” From our sample, it seems accurate to state that supply chains in the automotive industry are remarkably cooperative environments. Nevertheless, these numbers are especially notable if we consider that only 21.4% of the companies which said to cooperate on one of the levels with supply chain partners currently apply SCM software. It seems, thus, that the market for SCM software in the European automotive industry is not saturated yet. Why cooperating companies do not apply SCM software for cooperative processes yet is addressed in the following section.

3.2.4.2 Collaborative Planning As already mentioned in Chapter 2, cooperative planning is the core idea of SCM and, thus, those companies performing collaborative planning would be potential users of SCM software. The next figure illustrates which degree of collaboration participating companies have. For this

82

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

purpose, the survey provides three degrees of cooperation in analogy to the Wyner and Malone (1996) approach (see Section 2.3.4): •

Collaborative planning process with shared data. This is the type of cooperation in which partners agree on sharing planning methods or in surrendering the power of decision making to other planning parties. Of course, this category requires the sharing of relevant data for the planning.

•

Individual planning process with shared data. In this case, each party plans individually using its own planning models and methods. Planning is done through the use of their own data and shared data from cooperating partners that are relevant to the planning problem.

•

Individual planning process with individual data. This corresponds to the non-cooperative, decentralized planning in the Wyner and Malone (1996) approach. Data from other partners is not taken into account, nor is the planning structure adapted to other partners in the supply chain, nor is the power of decision shared.

21.9 29.8 34.3 36.5 37.6 38.2 38.8 39.9 43.8

transportation planning waste disposal/reflux planning SC monitoring sales/demand planning purchasing supply network/location planning SC controlling production planning inventory planning

N= 178

0

10

20

34.8 19.7 21.3

19.1 24.2 14.0 36.5 12.4 32.0 31.5 13.5 18.5 23.6 15.2 23.6 20.2 9.0 32.6 18.5 9.0 33.7 26.4 12.4 21.3 22.5 10.1 23.6

30

40

50

60

70

80

90

100

%

collaborative planning process with shared data

individual planning process with shared data

individual planning process with individual data

undeclared

Figure 23: Degree of Collaboration in Planning Processes

Inventory planning seems to be the task where most companies cooperate with their supply chain partners (43.8%) and where SCM software could provide most advantages. Strategic concepts in the field of inventory management such as VMI48 and CMI49 which have gained widespread

48

In a VMI system, the manufacturer is responsible for replenishing all merchandise and not only standard merchandise. Under this approach “suppliers generate orders based on mutually agreed upon objectives for inventory levels, fill rates and transaction costs, and demand information sent by their distributor customers. In this process, the buying function moves from the distributor back to the supplier, who takes over responsibility for placing orders. The distributor sends sales and inventory data to the supplier on a pre-arranged schedule typically, daily - and the VMI system determines what should be ordered based on the criteria the supplier and distributor have established. The supplier monitors the inventory status information to make sure that the distributor always has the appropriate amount of stock on hand when needed. The distributor can override the system when necessary, for example, if they anticipate an increased demand in the market” (Hall, 2002).

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

83

acceptance can provide an explanation for this number. Furthermore, the advanced supplier integration in the automotive industry (e.g. JIT strategies50) which demands transparent inventories throughout the supply chain can also be an explanation for the collaboration in this field. On the other hand, the planning of transportation processes seems to be the most individual planning task in our sample (21.9%). This can be the result of the fact that transportation is often outsourced to external logistics service providers.51 In accordance with this assumption, transportation is the individual planning activity with the highest amount of shared data (34.8%). For instance, Bosch Eisenach GmbH has outsourced all of its outbound logistics to the Paul Günther Industrielogistik GmbH (Buxmann et al., 2004, p. 163). The LSP individually plans the transportation, but requires detailed data on production and inventory processes from Bosch Noticeable is also the fact that sales and demand planning is the planning task that is most often performed using shared data (68%). The sharing of sales data for better forecasting at all levels of the supply chain is one of the factors that influence the bullwhip effect (see Section 2.2.4.2). In addition, the close relationship needed to realize supply concepts such as JIT require a certain degree of synchronization of production processes throughout the supply chain. This must be done, among other measures, through a propagation of the underlying forecasted sales numbers to all preceding production instances (the suppliers of components, modules, and systems) in order that they avoid inventory overheads or supply shortages which will lead to reduced service levels or higher costs in the supply chain. The fact that 95.5% of all participating companies perform at least one task in cooperation with supply chain partners (see Section 3.2.4.1), but only 41% perform at least one collaborative planning task, leads to the assumption that SC partners are far from opening up the full range of synergies which SCM software could support. The next section provides an insight into the actual levels of SCM software use in the sampled European Automotive Industry.

49

CMI is defined by the Collaborative Planning, Forecasting and Replenishment Committee as follows: “A form of Continuous Replenishment (CRP) in which the manufacturer is responsible for the replenishment of standard merchandise, while the retailer manages the replenishment of promotion merchandise” (CPFR, n.d., p. 1).

50

“An inventory control system that controls material flow into assembly and manufacturing plants by coordinating demand and supply to the point where desired materials arrive just in time for use. An inventory reduction strategy that feeds production lines with products delivered ‘just in time’. Developed by the auto industry, it refers to shipping goods in smaller, more frequent lots” (Vitasek, 2002).

51

This corresponds to one of the findings of the survey that revealed that 66% of the participants declared to outsource their shipments.

84

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

3.2.5 Supply Chain Management Software in the European Automotive Industry This section examines the status quo of SCM software use in the European automotive industry. In a first step, the diffusion rate of SCM solutions is analyzed (Section 3.2.5.1). Next, the goals of implementing SCM software are the focus of attention (Section 3.2.5.2). Section 3.2.5.3 deals with the benefits companies have derived from using these software solutions. Finally, Section 3.2.5.4 analyzes potential network effects in the SCM software market.

3.2.5.1 The Status Quo of Using Supply Chain Management Software First, the diffusion rate of SCM software was analyzed. We asked the participants if they had a SCM solution implemented, or if the implementation was still ongoing or at least planned.

Figure 24: Does Your Company Have a SCM Software Solution?

Figure 24 shows that 20.2% of the companies used SCM software solutions at the time; 14% were currently running an implementation project while 14.6% planned to implement a software solution in the future. Overall, 117 companies of our sample neither use nor implement SCM software. If one compares these numbers with the fact that 95.5% of the companies cooperate on at least one level (see Section 3.2.4.1), and 41% perform at least one cooperative planning task (see Section 3.2.4.2), one possible interpretation is that there is a significant gap between the need of software support for collaborative tasks in the SC and the actual support capabilities of current SCM software solutions. Furthermore, we asked the 61 companies that use or implement SCM software to name the chosen software solution. In our sample, the most common solutions were SAP (23%), Oracle (13.1%), i2 (11.5%), and Baan (8.2%). A fairly large percentage of companies chose to develop individual software solutions (26.2%) while more than one third used various other SCM software products. The fact that most companies in our sample chose to develop their own SCM solution may also corroborate the interpretation on Figure 24 which was formulated above.

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

85

As a next step, we examined the reasons why a big portion of companies still refrains from implementing SCM software (Figure 25).

benefit not quantified unnecessary implementation too expensive software too expensive no suitable solution available N = 117

0%

10%

20%

30%

40%

50%

60%

70%

Figure 25: Why Does Your Company not Use SCM Software? (Multiple Answers Were Allowed)

As Figure 25 shows, the main reason to refrain from implementing SCM software is that companies can hardly quantify its benefit for their organization. The statistical significance of this result is confirmed by a chi-square test at a level of .001. About 16.2% believe that SCM software is not necessary for them. Only a percentage of 10.3 consider the high software costs as the main reason. In addition, 11.1% expect an implementation process to be too costly. 8.5% of the companies did not find a suitable solution on the market. This is a remarkable finding for this sample since SCM software vendors often redesign their pricing strategies to enhance demand. These results indicate that although there are studies that reveal an important potential of the use of SCM software for reducing costs and improving efficiency (for an overview see for example Seidl, 2004, p. 181), potential customers still seem to demand a context-related quantification of their benefits in relation to a possible implementation of SCM software. Simulations of the customer context could probably be a better strategy for SCM software vendors in order to achieve higher product demand. On the one hand, 16.2% of the companies do not need SCM software at all which can mean that there are certain cooperative tasks in the European automotive industry that do not require support from dedicated SCM software. This might be corroborated by the fact that ERP software provides SCME capabilities as pointed out above. In this context, the survey required participating companies to declare how the planning tasks were supported by ICT (see Appendix D.6, Figure 60). The results indicate that some tasks which could be subscribed to the standard functionality of SCM software (Zeier, 2002, p. 3, pp. 8f) are

86

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

often performed by ERP software, or by individual developments within the company. It seems that there is a certain cannibalization between ERP software and SCM software due to the overlap of standard functionality and past customizing efforts. These results emphasize a well known and hard-to-solve problem: In the context of an economic analysis of ICT, it is generally more difficult to evaluate the potential benefits than to estimate the costs of an implementation. Given the problems of quantifying the benefits of SCM software, it is of particular interest to examine the goals companies seek to achieve with an implementation.

3.2.5.2 Goals of Implementing Supply Chain Management Software Section 2.2.3 presented the categories of goals of SCM as described in the literature. Assuming that SCM software provides an instrument for supporting and realizing SCM concepts, we asked the participants to evaluate these goals as “very important,” “important,” “not important,” or “irrelevant” for their decision on implementing SCM solutions. Figure 26 shows that 85.2% of the companies consider it either as very important, or important to achieve reductions in terms of inventory and shortfall. The second and third most important goals are reduction of lead time and production costs.

inventory and shortfall lead time production costs purchasing costs transportation costs 0%

very important

20%

40%

60%

80%

important

not important

irrelevant

not specified

100% N = 61

Figure 26: Persecuted Goals – Reductions

More than half of the questioned companies evaluate the goal of reducing purchasing costs to be very important, or important. Roughly 50% of the companies attempt to realize transportation cost reductions as one of their very important, or important goals. However, almost 40% of the companies consider transportation cost reductions as either not important or irrelevant. As already mentioned, this is probably due to the fact that many companies outsource their

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

87

shipments. The expected coordination and monitoring capabilities of SCM software that affect inventories and lead times seem to be the most important drive for companies in this sample. These two goals, and also the third most important (reduction of production costs) goal, are closely related to the degree of integration of business partners throughout the supply chain. Inventories and shortfalls are closely related to the transparency in the supply chain (refer to the bullwhip effect in Section 2.2.4.2). Lead times are also closely related to the efficient exchange of information in the supply chain (e.g. customer orders are notified to all members of the value chain in order to react as quickly as possible to shortages and to avoid waiting times). Furthermore, the synchronization of production efforts throughout the supply chain can lead to significant cost reductions (e.g. when implementing a JIT delivery strategy). The goals, in terms of improvements, are shown in Figure 27. We found that 72.1% of the companies rate service level improvements as a very important, or important goal in implementing SCM software. The second most important goal concerning improvements according to our survey is the redesign of the supply chain. In addition, almost 46% of the companies evaluate it as very important or important to realize improvements concerning supplier evaluation and selection.

service level supply chain redesign supplier evaluation and selection cooperation

0%

20%

40%

60%

80%

not important

irrelevant

not specified

100% N = 61

very important

important

Figure 27: Persecuted Goals – Improvements

The goal that was ranked least important by the participating companies is the improvement of cooperation: Only 27.8% of the companies evaluate it as either very important, or important. This corresponds to the findings of case studies showing that SCM software is used mostly on an intra-organizational level today (Buxmann & König, 2000, pp. 159ff; Buxmann et al., 2004, pp. 141ff). It is interesting to note that companies are dependent upon each other as parts of a supply chain and achieve competitive advantages by cooperating (Whipple & Gentry, 2000; Helms et al.,

88

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

2000). In addition, it is widely accepted that cooperation in supply chains needs to be supported by corresponding software solutions (Chu et al., 2002). However, the cooperative potential of SCM software does not seem to be leveraged yet. In this context, the companies were asked which were the challenges of an inter-organizational use of SCM software. The answers revealed that 50% of the companies consider the unclear cost/benefit-ratio as a main challenge in using SCM software on an inter-organizational level (see Appendix D.6, Figure 61). The assessment of cooperation benefits seems, thus, to be the main challenge for an inter-organizational use of SCM software. If one assumes that SCM software is a valid instrument for supporting SCM activities, it can be assumed that one demanding challenge for research activities in the field of Information Systems is the assessment of benefits realized by cooperation in supply chains with the use of ICT.

3.2.5.3 Evaluation of Benefits from Using Supply Chain Management Software One approach for assessing the benefits of cooperation can be an empirical study. In this section, we examine the results companies have achieved through SCM software use (provided by a nonprobability sample). Some of the 36 companies in our sample using SCM software solutions did not quantify all of the benefits they achieved. In particular, those benefits that were of less importance to the companies were not quantified in many cases. Hence, one has to be careful when interpreting these results. However, they may provide an insight into potential benefits and, thus, can be of particular interest for both companies who are still deciding whether to implement SCM software and software vendors.

3

inventory and shortfall reductions lead time reductions

12

2

6

13

6

10

transportation costs reductions

4

2

production costs reductions

6

3

purchasing costs reductions

8

6

3

service level improvements

4

0-5%

4

6-15%

7

16-25%

1

18

19

17

4

> 25%

13

18

2

8

11

4

2

not quantified

N = 36

Figure 28: Evaluation of Supply Chain Management Software – I (Shown in Absolute Figures)

Figure 28 reveals that 3 out of 36 companies achieved an inventory or shortfall reduction by 0% to 5%. 12 companies were able to reduce their inventory and shortfall by 6% to 15% while six

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

89

companies reduced their inventory levels up to 25%. Four companies realized tremendous reductions and indicated that their inventory and shortfall level went down by more than 25%. The results concerning lead time reductions are similar. In addition, eight out of 36 companies reduced their production costs by 0% to 5%, six companies managed to cut their costs by 6% to 15%. Similar results can be found for purchasing and transportation costs reductions. 17 out of 36 companies were able to improve their service level; in four cases even by more than 25% while the largest group (7) achieved improvements between 16% and 25%. In spite of the small number of companies that provided information about this topic, the results indicate that using SCM software can help achieving cost and time reductions as well as service level improvements: The companies that quantified the benefits in average reduced their inventory and shortfalls by 17.8% and their lead time by 14.8%. Moreover, the transportation costs were reduced by an average of 10.2%, the production costs by 10.1%, and the purchasing costs by 8.5%. Finally, the service level in average could be improved by 20.2%. Improvements in cooperation, supply chain redesign as well as supplier evaluation and selection are even harder to quantify. Thus, the participants were asked to rate the results they achieved in this respect as “very good,” “good,” “average,” “bad,” or “very bad.” The results in Figure 29 show that 31 out of 36 companies improved their cooperation by an average, or above average degree. None of the questioned companies indicated that in regard to cooperation their software performed bad, or very bad. 11 companies label their SCM software as good, and 12 even find it to be very good. All in all, these results show that it is not the bad performance of SCM software in this regard that bars companies from cooperating. In contrast, SCM software use appears to improve cooperation in supply chains remarkably.

12

cooperation

supply chain redesign supplier evaluation and selection

11

3

14

5

8

14

7

11

5

0

5

0

6

0

7

N = 36

very good

good

average

bad

very bad

not specified

Figure 29: Evaluation of Supply Chain Management Software – II (Shown in Absolute Values)

90

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

Only three companies rate their improvements in terms of supply chain redesign as very good; 14 participants evaluate their improvements as good. 14 companies were able to improve their supply chain design on an average level so far. If we review the results for supplier evaluation and selection, we find that five companies evaluate their solution as very good, and seven as good. However, six companies find their solution to work poorly in this respect. In addition to the benefits of using SCM software, the interaction and compatibility with other systems within the company (especially ERP systems) and among the business partners can be relevant for deciding whether to implement SCM software or not. Therefore, the next section outlines the network effects that might play a role for the participants of our survey in this respect.

3.2.5.4 Supply Chain Management Software and Network Effects in the European Automotive Industry We can observe that traditional supply chains are being increasingly displaced by network-like structures (Buxmann et al., 2004, p. 5). For example, a company can assume an active role in several supply chains at once; a manufacturer can purchase materials from different suppliers by convoking reverse auctions (Gebauer & Buxmann, 2000). The integrative idea of SCM and the underlying network of companies involved in SCM processes can, thus, make it necessary to take network effects into account when, for example, evaluating the use of SCM software. This section presents additional results of the empirical study that will be interpreted in light of the network effect theory (Farrel & Saloner, 1985; Katz & Shapiro, 1985; Bensen & Farrel, 1994; Weitzel et al., 2000; Buxmann, 2001a). When a company evaluates SCM software use at the inter-organizational level, so-called network effects have to be taken into account. The network effect theory describes the utility of some goods to be dependent on the number of people using these goods. The network effects that affect this utility can, thus, be divided into direct and indirect network effects (Katz & Shapiro, 1985, p. 424): •

Direct network effects. They result from the compatibility between system elements. For example, direct network effects let users of an email system benefit from a high diffusion rate of this system. The more users apply the email system, the higher will be the experienced utility for each user. In this case, the quality of the integrated planning with a certain SCM software solution will increase with the number of business partners in the supply chain using the same SCM software. The integration of the required planning data

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

91

and planning results will be easier because of similar data structures and common exchange formats. •

Indirect network effects. They describe the positive dependence between the spread of a certain good and the availability of complementary goods. For example, users of an operating system like WindowsTM experience higher indirect network effects than users of the operating system BeOSTM. There is a much broader supply for compatible software for the Windows operating system which might result in a higher utility. In the context of SCM software, the spread of a certain ERP software may influence the spread of the SCM software solution from the same vendor. Since current SCM software solutions rely on the underlying ERP software to execute the planned tasks, the integration between both systems plays an important role. This integration is easier if both systems are from the same software vendor.

On the other hand, we have to make a distinction between the utility derived from these network effects, and the utility of a good that is not influenced by network effects: the basic utility. For instance, a spreadsheet program has a basic utility derived from its functionality that allows the user to perform calculations and data analyses. In addition, the network effect utility results from the ease of exchanging files with other software users (direct network effect) as well as from the availability of online resources and plug-ins (indirect network effect). In this context, an instant messaging system would be the kind of good which primarily offers only network effect utility while a screen-saver would be just providing basic utility. In the field of SCM software, the basic utility of the software would be the functionality that allows the company to plan logistics activities and to support supply chain processes. On the other hand, the network effect utility would be derived from the diffusion of the same (or compatible) SCM software solution in the supply chain. A broad diffusion of the same product will result in an easier integration of data from business partners, exchange of planning results, and adoption of planning methods supported by the software. For every actor of the supply chain, the network effect factor depends on both the level of interaction (in terms of frequency of communication with partners), and the level of integration (in terms of the amount of shared information). The more the logistics processes of a company are dependent on the interaction and integration with business partners, the greater the advantages of having compatible SCM software solutions. In order to see how relevant network effects might be in the context of SCM software, we conducted specific questions in our survey.

92

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

Concerning the selection criteria for the particular SCM software solutions, we asked the participants of our survey to evaluate different criteria as “very important,” “important,” “not important,” or “irrelevant” (Figure 30). The results show that functionality and price as well as compatibility within the company play key roles in the decision-making process: More than 80% of the companies evaluate these factors as being very important, or important selection criteria. By comparison, criteria relating to network effect utility, like compatibility to business partners as well as the diffusion rate of a software solution, are considered to be less important. Thus, the selection of SCM software seems to be influenced more by the requirements of intra-organizational processes.

cos ts tatio n me n le p m /i pr ic e ty on ali func ti

it ati bil c omp

hin y wit

any com p

ate ion r diffu s cte d ex pe n ers s p art s ines to bu y it il ati b c omp n rate fus io nt d if c urre ow ow-h ng k n ex is ti ts n c os cati o g/edu in in tra 0%

20%

40%

60%

80%

100% N = 61

very important

important

not important

irrelevant

not specified

Figure 30: Selection Criteria for SCM Software Solutions

Still, 13.1% of the companies find the current diffusion rate for SCM software solutions they purchase to be very important, and 41% to be important. This confirms that, among other reasons, the installed base of a SCM solution influences the selection process and may emphasize the relevance of network effects in this market. Of course, the diffusion rate can also be understood as a sign of quality, for example; and therefore, be considered of importance to companies. In addition, 68.9% of the questioned companies considered it very important, or important that the selected solution has an expected high diffusion rate. This can be interpreted as the expectation of a firm to benefit from a software product that has a certain market share since it will, more likely, be compatible and allow the company to benefit from possible network

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

93

effects (direct network effects, e.g. ease of information sharing, and indirect network effects, e.g. support, add-ons, consulting services, etc.). It seems that the basic utility of SCM is more important than the network utility. Still, the network utility plays an important role in the selection process of suitable SCM software. Table 13: Dependencies ERP-Supply Chain Management Software Solution52 SAP

57 companies have implemented SAP (R/3 or R/2) ERP software solutions. Out of these, 21 companies use a Supply Chain Management software solution

SCM software vendor

Amount of implementing firms

SAP

14

Individual software

6

Baan

4

i2

3

Oracle

2

Brain

1

J.D. Edwards

1

Oracle

14 companies have implemented Oracle ERP software solutions. Out of these, 10 companies use a Supply Chain Management software solution

SCM software vendor

Amount of implementing firms

Oracle

7

SAP

2

Baan

2

i2

2

Individual software

1

Baan

11 companies have implemented Baan ERP software solutions. Out of these, 7 companies use a Supply Chain Management software solution

SCM software vendor

Amount of implementing firms

Baan

5

SAP

3

i2

1

In addition, it appears that companies tend to choose the SCM software from the vendor that supplies the in-house ERP system.53 As Table 13 shows, 21 companies of our sample have

52

As companies may use several SCM software solutions, multiple answers were allowed.

53

However, this finding can not be confirmed statistically because of the small number of companies in our sample using specific software solutions.

94

3. Cooperation in Supply Chains and SCM Software Use in the European Automotive Industry

implemented SAP’s ERP software solution and are also using SAP SCM software. Two thirds (14 companies) of the companies use SAP APO while only 6 of the companies decided to develop their individual software solution, 4 companies chose to complement SAP’s ERP software with Baan’s SCM software solution, 3 companies use i2, and 2 companies apply Oracle SCM software. A similar situation can be found in other ERP solutions: For each ERP software solution it seems that the largest percentage chose to implement the SCM software solution offered by the respective ERP vendor. Likewise, the companies with an individual ERP software solution developed their own individual SCM software as well. These findings again indicate that compatibility within the companies seems to be more important than inter-organizational compatibility and cooperation. Taking into account the small size of this sample, Table 13 appears to show indirect network effects that influence the purchasing decision of SCM software. This means that the diffusion rate of a certain ERP software affects the expected diffusion rate of a complementary product such as SCM software solutions. Indirect network effects seem, thus, to play an important role in the SCM software market. The indirect network effects can, for example, be explained by the easier integration of both types of software.

3.2.6 Summary of Results Given the attention SCM software has achieved over the last years, it appears of particular interest to investigate the potential benefits as well as problems of implementing such software solutions. The results indicate a gaining importance of SCM software. However, there are still some serious barriers to a further distribution. However, the high costs are not the main reason why companies refrain from implementing SCM software. It is rather the unquantified benefit that causes a rejection of its implementation. Considering these findings, one major goal of our survey was to quantify cost reductions as well as improvements achieved by implementing SCM software. For example, the survey revealed that the participating companies reduced, on average, their inventory shortfalls by 17.8%, and their lead time by 14.8%. Moreover, the transportation costs were reduced, on average, by 10.2%, the production costs by 10.1%, and the purchasing costs by 8.5%. Additionally, the service level, on average, could be improved by 20.2%.

3.2 SCM Software as an Instrument for Cooperative Planning in Supply Chains

95

However, the cooperative potential of SCM software is not leveraged yet: Improvement in cooperation is considered to be less important than other goals. Thus, companies rank the unclear cost/benefit-ratio as the major problem of an inter-organizational use for these software solutions. Therefore, one demanding challenge for research activities is the assessment of benefits realized by cooperation in supply chains. For example, concerning demand and inventory planning some advanced approaches exist already (Lee et al., 1997a). For this purpose, Chapter 5 presents a prototype for evaluating the effects of cooperative planning in supply chains. The objective of this prototype is to estimate the benefits of cooperative planning scenarios versus the results of non-cooperative planning. This added value of cooperation could serve as a basis for the decision as to whether implement SCM software or not. The survey also shows that, although the basic utility of the software is still the key selection parameter, both direct and indirect network effects assume an increasingly important role in deciding if and what SCM solution a company is to implement. The compatibility with business partners and the current as well as the expected diffusion rate of a SCM solution are important factors which influence the purchasing process. Moreover, the installed ERP software affects the decision of which SCM software to purchase. Therefore, most companies select the SCM solution of the vendor that provides their current ERP software. These considerations may affect the way SCM software vendors address their target group. In light of network effects, the leading share of SAP in the ERP market (33.4% in our sample) leads one to assume that the SAP SCM software solution (APO) will not only become but also remain the leading SCM software vendor if the current ERP market situation remains as is.

4 The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

This chapter presents an analysis of the supply chain of the Audi A8 V8 4.0l diesel engine with two main goals: •

First, gathering information to quantify the bullwhip effect contained in this simple supply chain and

•

second, providing the basis for evaluation of the cooperation scenarios identified in Section 3.1.2 to determine if they could be implemented in this engine supply chain and what advantages could be gained.

For this purpose, this dissertation uses an empirical approach in form of a qualitative case study (Benbasat et al., 1987; Lee, 1989; Yin, 2002; Dubé & Paré, 2003). The reasons for using a qualitative approach in this evaluation include the assumption that the relevant variables of the problems of cooperation are context driven and that the internal dynamics, implementation, and quality need to be understood as well (Creswell, 1994; Patton, 1987). Since this dissertation follows the interpretive approach rather than the positive approach, this case study is not intended to provide repeatability, but to gain a deeper understanding of the phenomenon cooperation in supply chains of the European automotive industry (Darke et al., 1998, p. 277). The author acknowledges the subjectivity of this process since this research attempts to “understand phenomena through accessing the meanings that participants assign to them” (Orlikowski & Baroudi, 1991, p. 5). Nevertheless, some quantitative methods are also involved in this case study (see Sections 4.2.2, 4.2.3, and 4.3.1) and some positive analysis (Darke et al., 1998, p. 276) is realized by the examination of product orders, deliveries, and inventories throughout this supply chain.

98

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

This chapter first presents all involved companies briefly. The supply chain of the Audi A8 V8 4.0l diesel engine is described in detail in Section 4.2. Section 4.3 attempts to determine the bullwhip effect by considering the data gathered in the case study (Section 4.3.1). This section provides also an evaluation of the benefits involved with the implementation of cooperation scenarios that were identified in Chapter 3 in this particular supply chain (Section 4.3.2). Further, Section 4.3.3 discusses the potential benefits of the implementation of a monitoring system for this particular supply chain. The chapter ends with the summary of the findings of the case study (Section 4.4).

4.1 Exposés of Companies in the Supply Chain of the V8 4.0l Diesel Engine In this section, some general information on Audi is presented, more specifically on the company’s history and its financial situation. Thereafter, a rough sketch of Audi’s up- and downstream supply chain will be provided. Exposés of the companies involved in the case study are also included.

4.1.1 Audi AG 4.1.1.1 General Overview of Audi AG Audi is an internationally renowned manufacturer of high-quality cars and is incorporated in Germany. The company is the result of two mergers. The first merger occurred in 1932 when the four automobile manufacturers Audi, DKW, Horch, and Wanderer merged to form Auto Union AG. To symbolize this union, the company adopted a logo that is made up of four intertwined rings – this sign is still used today to represent Audi AG (Audi AG, n.d. a). In 1969, the second merger was completed, when Auto Union AG and NSU joint forces to form Audi NSU Auto Union AG which was renamed Audi AG in 1985 (Audi AG, n.d. b). Since 1964, the company is a fully owned subsidiary of Volkswagenwerk AG, today known as Volkswagen AG (Audi AG, n.d. b). About 99 percent of Audi’s share capital is held by Volkswagen AG (Audi AG, 2005). Over the years, Audi acquired shares of several companies and also founded a number of subsidiaries. The result is Audi Group with locations all over the world. As shown in Figure 31, Audi Group is comprised of Audi Hungaria Motor Kft. (Audi Hungaria), Cosworth Technology Limited, Lamborghini Group, Autogerma S.p.A., Quattro GmbH, Audi do Brasil e Cia., Audi Senna Ltda., and Audi AG.

4.1 Exposés of Companies in the Supply Chain of the V8 4.0l Diesel Engine

Cosworth Technology Northampton Wellingborough Audi Hungaria Worcester Cosworth Technology Györ Novi AUDI AG Ingolstadt Lamborghini Neckarsulm Group Sant‘Agata Bolognese Autogerma Audi Senna Verona São Paulo Audi do Brasil Curitiba

99

FAW-Volkswagen Changchun

Figure 31: Worldwide Locations of Audi Group (Source: Audi AG, 2003)

During the past nine financial years Audi Group was very successful, and one record year followed another (Audi AG, 2005). Table 14 gives an overview of key figures for the years 2003 and 2004. Table 14: Key Figures of Audi Group for 2003 and 2004 Audi Group Vehicle sales Audi (total) Germany Outside Germany Lamborghini Other VW Group brands Revenue Profit from ordinary business activities (Profit before tax and transfers54) Net profit

31 Dec. 2004 971,832 779,441 235,092 544,349 1,592 83,372 24,506,000,000

31 Dec. 2003 1,003,791 769,893 237,786 532,107 1,305 76,121 23,406,000,000

Change in % -3.2 1.2 -1.1 2.3 22.0 -18.0 4.7

1,142,000,000

1,101,000,000

3.7

871

811

7.4

(Source: Based on Audi AG, 2005)

As shown in Table 14, Audi Group was able to increase its revenue in 2004 by 4.7 percent (compared to 2003) to EUR 24,503 million. In addition, the group’s profit before and after taxes were also increased considerably. Interestingly, the growth in revenues is not supported by unit sales of Audi branded cars which decreased by 3.2 percent. (see Table 14). Unit sales of Audi branded vehicles reached 971,832 (see Table 14) which means that the previous record from 2003 could not be topped. On the other hand, Table 15 shows that the number of vehicles produced also increased from 2003 to 2004. 54

By agreement between Audi and VW, a proportion of the profit is to be transferred to VW (AUDI AG, 2003d).

100

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

Table 15: Production of Vehicles for Audi Group in 2003 and 2004 Production of vehicles

2004

2003

Audi A2

19,745

27,323

Audi A3

174,750

151,117

Audi A4

313,027

327,463

31,962

29,285

-

-

Audi A4 Cabriolet Audi RS 4 Audi RS 6 Audi TT Coupé Audi TT Roadster

1,233

4,841

14,753

20,807

8,852

11,530

181,701

148,477

Audi allroad Quattro

14,842

17,634

Audi A8

22,429

21,748

783,294

760,255

1,294

933

Audi A6

Total, Audi brand Lamborghini Gallardo Lamborghini Murciélago Total, Lamborghini brand Total, group

384

424

1,678

1,357

784,972

761,582

(Source: Based on Audi AG, 2005)

Table 15 not only grants an overview of unit production of the Audi Group, but it also shows the product line of both Audi Group brands (Audi and Lamborghini). Most cars of the Audi brand are manufactured in Ingolstadt (a total of 528,739 vehicles in 2004). In this plant, the lines A3, A4, A4 Avant, S3, S4, and S4 Avant are produced. The Audi TT is partly produced in Ingolstadt and partly in Hungary (together with Audi Hungaria). At Audi’s second German plant, Neckarsulm, 239,950 vehicles were produced in 2004. They were from the following product lines: A2, A6, A6 Avant, allroad quattro, S6, S6 Avant, RS6, RS6 Avant, and A8/A8 longwheelbase (Audi AG, 2005). The strong numbers in unit sales, which are also reflected in production numbers, are underscored by the development of Audi’s market shares in different countries around the world. Table 16 shows that Audi’s home market, i.e. the Germany market, is its most important one: In 2004, 30.16 percent of Audi branded cars were sold in Germany. Here, the company also holds its largest market share with 7.2 percent. The United States (US) are the second most important market for Audi regarding unit sales; however, the company does not hold a very strong market share in this country (only 1.8 percent among foreign brands). The third most important market for Audi is Great Britain.

4.1 Exposés of Companies in the Supply Chain of the V8 4.0l Diesel Engine

101

From Table 16 one can conclude that Europe is by far the most important region for Audi, followed by the North American market including the US and Canada. The Asian-Pacific market is not as strong when compared to Europe and North America, but it seems to have an enormous growth potential. For example, the Chinese automobile market grew by 15.4 percent compared to the previous year. Moreover, Audi experienced its strongest growth in terms of unit sales from 2003 to 2004 in South Africa (Audi AG, 2005). Table 16: Audi AG’s Position in Major Markets Unit sales 2004

Year-per-year percentage change

Market share in 2004 in %

Year-per-year percentage change, overall market

235,092

-1.1

7.2

0.9

USA

77,917

-9.8

1.855

2.2

Great Britain

77,882

11.1

3.0

-0.5

Italy

50,500

-1.6

2.4

0.8

Spain (including Canary Islands)

43,764

6.4

3.0

9.7

France

37,676

0.6

1.8

0.2

China (including Hong Kong)

64,018

0.8

2.4

15.4

Belgium

21,509

11.9

4.4

5.7

Switzerland

14,105

-6.6

5.2

-1.2

Austria

15,711

-7.2

4.9

3.7

Netherlands

15,038

5.4

2.3

-1.0

Sweden

11,598

-0.4

4.2

1.2

Japan

13,751

4.7

5.7

-1.6

South Africa

10,086

28.9

3.0

27.7

7,422

-5.6

0.9

-5.2

779,441

1.2

1,592

22.0

Germany

Canada Audi, worldwide Lamborghini, worldwide

(Source: Based on Audi AG, 2005)

4.1.1.2 Supply Chain Related Overview of Audi AG Audi branded vehicles are mostly produced in Ingolstadt and Neckarsulm, Germany (see 4.1.1.1). Other car production plants are located in Györ (Hungary), Changchun (China), Curitiba (Brazil), and Bangkok (Thailand) (Audi AG, n.d. c). Additionally, there are plants that only produce parts but not entire cars, e.g. Györ in Hungary producing engines (Audi AG, n.d. d). Another type of

55

Only foreign brands were considered in the calculation of this market share.

102

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

plant is the so-called CKD56 plant (e.g. in Pealing Jaya, Selangor Darul Ehsan, Malaysia) where vehicles are not manufactured but only assembled from CKD kits which are delivered, for example, from Ingolstadt, Germany (Audi AG, n.d. e). In order to build its cars, Audi does not manufacture most of the parts, systems, and modules internally,57 but buys them from 1,000-1,100 suppliers (Thomas Müller, personal communication, June 26, 2003) which are located all around the world (Werner Diedrich, personal communication, July 02, 2003). Some of these suppliers are just-in-time, or just-in-sequence suppliers. Goods delivered by these suppliers are not stored in an Audi warehouse and are not quality checked by Audi before they are used but are transported directly to the production line and immediately built in (Thomas Müller, personal communication, June 26, 2003). Just-in-time and just-in-sequence delivery is used for parts characterized by at least one of the following: high complexity, high number of variants, and bulkiness (Thomas Müller, personal communication, June 26, 2003). JIT and sequenced delivery are used for an increasing portion of parts because parts are getting more complex and also the number of variants increases. The alternative to JIT or sequenced delivery would be to have several containers standing at the production line, and each containing one of the possible variants. However, this can hardly work because, firstly, there is not enough space available; and, second, it would be almost impossible for workers to know which part is the correct one because it is often hard to tell the difference between variants (e.g. if colors differ just slightly) (Thomas Müller, personal communication, June 26, 2003). Most of the just-in-time and just-in-sequence suppliers to Audi Ingolstadt are located in a complex called Güterverteilzentrum (GVZ) which is located across the street from Audi and which is where just-in-time and just-in-sequence suppliers assemble their parts. Exceptions are Faurecia (delivering seats) and Peguform GmbH & Co. KG (delivering door trims). Both are located in Neuburg an der Donau (Thomas Müller, personal communication, June 26, 2003) which is about 27 kilometers west of Ingolstadt. Many just-in-time and just-in-sequence suppliers of Audi Neckarsulm are located in the Gewerbe- und Industriepark Friedrichshall (GIF) which serves the same function as the GVZ in Ingolstadt (Alfred Raible, personal communication, July 04, 2003).

56

CKD stands for “completely knocked down”. It means that a car is sent in individual parts or in modules to its country of destination where it is then assembled (AUDI AG, 2003f).

57

Hereafter, the word “part” will also be used to refer to modules and systems.

4.1 Exposés of Companies in the Supply Chain of the V8 4.0l Diesel Engine

103

When parts are delivered just in sequence, they are not put into storage at Audi and there is no precise supply call-off [“Feinabruf” (FAB)]. Instead the supplier is automatically notified about which variant of its part to deliver for a certain car – this happens right at the moment at which the car is being produced. The amount of time that a supplier has to deliver its part varies depending on when the part is built into the car. For example, the supplier Dräxelmaier GmbH has only 50 minutes to assemble and to deliver its part (“Schalttafelleitungssatz”) whereas Faurecia as a seat supplier has about seven hours. The reason for this difference is that “Schalttafelleitungssätze” are built in the car at an earlier stage than seats (Thomas Müller, personal communication, June 26, 2003). The cars produced at the above mentioned locations are, in part, sold in the country where they are produced and, in part, are exported to be sold in different countries (see also Table 16). To manage the global distribution of Audi branded cars, the company’s sales unit is organized in five divisions: Asia/Pacific, Southern Europe/Africa/AGCC58, Northern/Middle/Eastern Europe, Northern/Southern America, and Germany. In some countries, distribution is managed by free importers, in others by subsidiaries of Audi AG (for example, in Italy by Autogerma S.p.A as a general importer), and in others by regional offices (for example, offices in Singapore and Beijing organize the distribution in the Asia/Pacific region, except for Japan) (Audi AG, 2003e). In Germany, there are roughly 2,000 independent dealers who are authorized to sell Audi branded vehicles based on a contractual agreement with Audi. Since the majority of dealers is independent, it follows that Audi is usually not directly involved in the distribution of its products to the end customer. Dealers purchase the product from Audi and then sell it to the end customer which means that the purchasing agreement is not made between Audi and the end customer, but between the end customer and the independent dealer (Thomas Hallatschek & Corinna Frese, personal communication, May 21, 2003). However, Audi allegedly is attempting to take over more outlets to gain more control (Autohaus-Online, 2003). The material and vehicle flow between suppliers and Audi’s plants as well as between Audi’s plants and dealers/importers is managed by VW Transport GmbH & Co. OHG (VW T) (Roland Scheidler, personal communication, April 30, 2003; Volkswagen Transport, n.d.), which is a wholly owned subsidiary of the Volkswagen Group (VW Group) and the principal transport contractor of the VW Group (Volkswagen Transport, n.d.). VW T does not conduct the actual transport of materials and vehicles, but the company manages the transport by functioning as a freight forwarder (see Section 3.1.1). As such, it purchases transportation services from select

58

The AGCC include the countries Bahrain, Kuwait, Qatar, Oman, Saudi Arabia, and United Arab Emirates.

104

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

partners to assure the supply of materials and the delivery of vehicles around the world using rail, truck, ship, and air as possible modes of transportation (Volkswagen Transport, n.d.). Usually, Audi pays for all transportation services (Roland Scheidler, personal communication, April 30, 2003).

engines

Rail transport between Gyor and Ingolstadt (both directions)

engine components

engines complete vehicles Ingolstadt

Neckarsulm

engines

Gyor

engine components painted bodies

engine components bought-in vehicle parts

engine components bought-in vehicle parts

regional carriers

suppliers Figure 32: Transportation between Audi in Györ and Ingolstadt/Neckarsulm as well as Other Plants (Source: Adapted from Audi Hungaria Motor Kft., n.d.)

4.1.2 Audi Hungaria Motor Kft. Audi Hungaria Motor Kft. is a fully owned subsidiary of Audi AG which has been founded in 1993 in order to manufacture engines as well as components and to build motor vehicles (currently Audi TT, A3) in Györ, Hungary (Audi Hungaria Motor Kft., n.d.). The plant has the capacity to produce about 6,880 engines per day (3,000 R4 Otto, 2,230 R4 TDI PD, 1,400 V6 Otto and TDI as well as 250 V8 Otto and TDI) which adds up to 1,480,630 engines per year. Additionally, it is able to manufacture 23,605 cars per year (Audi AG, 2005, Audi Hungaria Motor Kft., n.d.). As shown in Figure 32, the engines and vehicles produced in Györ are transported by rail to the Audi plant in Ingolstadt and from there, the engines are shipped to Neckarsulm as well to other plants located in Europe, Middle and Southern America, South Africa, and Asia. Not only Audi plants are served, but also sites run by VW, Seat S.A., and Škoda Auto a.s. Engine components are transported from these other plants located around the world to Ingolstadt and from there, they are railed to Györ together with painted vehicle bodies (Audi Hungaria Motor Kft., n.d.).

4.1 Exposés of Companies in the Supply Chain of the V8 4.0l Diesel Engine

105

The production of engines for the VW Group and of cars for Audi allowed Audi Hungaria to generate a sales return of EUR 3,552 million and EUR 280.8 million in earnings after taxes in 2002 (Audi Hungaria Motor Kft., n.d.).

4.1.3 TCG Unitech Systemtechnik TCG Systemtechnik is part of the Trident Components Group (TCG Group) and is considered the group’s power train competence center (TCG, n.d. a). The group runs seven manufacturing plants and two sales offices located across Europe. The TCG Systemtechnik branch is located in Micheldorf, Austria (TCG, 2001). The company’s product portfolio includes water pumps, oil pumps, front covers, camshaft timing systems, and cylinder head covers (TCG, n.d. b); its major customers are the large automobile OEMs, e.g. Audi, BMW AG, Daimler Chrysler AG, General Motors Corporation, Saab Automobile AB, and VW (TCG, n.d. c). In 2001, the group was able to generate EUR 259.1 million in sales and EUR 9.6 million in operating profit before exceptional items (TCG, 2001).

4.1.4 Gustav Wahler GmbH u. Co. KG Wahler’s headquarter is located in Esslingen near Stuttgart. It runs three manufacturing plants: one in Esslingen, one in Oberboihingen (also near Stuttgart), and one in Piracicaba, S.P. (Brazil) (Wahler, n.d. c). This company divides its products into the three categories “temperature control systems“ (cooling liquid, engine oil, transmission oil, diesel fuel, and intake air), “pneumatic actuators” (exhaust-gas return valves, pressure regulators, overflow valves, altitude regulating switches, positive crankcase ventilation valves), and “flexible metal pipes“ (exhaust-gas return pipes, oil return pipes, flexible metal bellows, expansion joints) (Wahler, n.d. a). Important customers for these products are OEMs and major suppliers, e.g. Audi, BMW AG, Robert Bosch GmbH, Dr. Ing. h.c. F. Porsche AG, Siemens AG, and VW (Wahler, n.d. b). In 2002, Wahler was able to generate EUR 160. million in sales.

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine In this section, the supply chain of the Audi A8 V8 4.0l diesel engine will be looked at in detail because this supply chain will serve as the basis for the latter evaluation of the advantages of cooperation scenarios in Section 3.1.2.

106

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

4.2.1 Description of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine Produced in Neckarsulm, the A8 is Audi’s current product for the upper segment of the car market and Audi offers a V8 4.0l diesel engine only for this model.59 Altogether, there are about 10,146 part numbers60 for the A8: 895 are manufactured in Neckarsulm and 9,251 are bought from circa 516 suppliers located around the world, and a supplier can either be another company outside the VW Group, or another VW Group plant (Martin Hackner, personal communication, July 03, 2003; Werner Diedrich, personal communication, July 03, 2003). The supply chain of this engine, which is visualized in Figure 33, has been chosen for this case study because it is a good example for demonstrating the bullwhip effect: Since there is only one customer for each part, and since each customer only has one source, one would assume that orders to Audi for A8s with this engine would be reflected in the orders by other members further upstream. However, as will be shown in Section 4.3.1, this is not the case. Figure 33 only shows the critical path61 of the supply chain for the V8 4.0l diesel engine.62 The reason for only looking at a selected path is that examining the entire supply chain of the engine would exceed all levels due to manageable complexity. As can be seen in Figure 33, the critical path is composed of four levels and companies: Audi Neckarsulm, Audi Hungaria, TCG Systemtechnik, and Gustav Wahler GmbH u. Co. KG (Wahler). Between all members of the engine supply chain there is a single sourcing relationship, and there are one-to-one relationships between all companies which means that one thermostat goes into one water pump, one water pump goes into one engine, and one engine goes into one car (Gisela Kastner, personal communication, July 04, 2003a; Jörg Binick, personal communication, July 07, 2003; Katalin Szorady, personal communication, July 16, 2003; Markus Bedic, personal communication, July 04, 2003).63

59

For a picture of the Audi A8, see Figure 3 in Appendix A.3; for a picture of this engine, see Figure 4 and Figure 5 in Appendix A.3.

60

One part can have numerous part numbers if there are different variants of that part. For example, if there are five colors for a seat, this seat has five part numbers so that each number represents one specific seat color.

61

The critical path must not be understood in strict terms as from the project management field of research. Towards the end of this section it will be explained how a critical path is determined in general, and in particular it will be described how the members of the engine supply chain have been selected.

62

This supply chain or rather the critical path will from now on be referred to as the engine chain or engine supply chain. For short exposés on the companies involved (except Audi Neckarsulm), see Sections 4.1.2, 4.1.2, and 4.1.4.

63

There is only one exception to the statement that there is a one-to-one relationship: TCG Systemtechnik not only delivers water pumps to Audi Hungaria, but also to VW in Kassel (the central spare part warehouse of the VW Group). However, VW in Kassel has not been included in the critical path by the responsible people and,

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine Tier 3

Tier 2

weekly supply call-off via fax for thermostats

Wahler plant locations: Esslingen (Germany)

OEM

Tier 1

weekly supply call-off via EDI for water pumps

TCG Systemtechnik plant location: Micheldorf (Austria)

107

weekly supply call-off (LAB) and daily precise supply call-offs (FAB) via EDI for V8 4.0l diesel engine

Audi Hungaria plant location: Györ (Hungaria)

Audi Neckarsulm plant location: Neckarsulm (Germany)

lead time: 3 days

lead time: 4 days

lead time: 2 days

weekly delivers: thermostats

weekly delivers: water pumps

daily delivers: V8 4.0l diesel engine

flow of products and materials flow of information

Figure 33: The Supply Chain of the V8 4.0l Diesel Engine (Source: Gisela Kastner, personal communication, July 04, 2003a; ICON, 2003d; Jörg Binick, personal communication, July 07, 2003; Katalin Szorady, personal communication, July 16, 2003; Markus Bedic, personal communication, July 04, 2003)

In Esslingen, Wahler produces thermostats for TCG Systemtechnik64 which uses these parts to manufacture water pumps for Audi Hungaria (Gisela Kastner, personal communication, July 04, 2003; Markus Bedic, personal communication, July 04, 2003). TCG Systemtechnik orders the thermostats at weekly intervals with a supply call-off via fax.65 In the engine supply chain, a supply call-off constitutes an order with which each supply call-off supersedes its predecessor. Such a supply call-off contains weekly demand information for the next six months, and the upcoming four weeks are legally binding while demand for the other twenty weeks can still be subject to change (Gisela Kastner, personal communication, July 04, 2003a; see also Figure 53). TCG Systemtechnik pursues a (t,S)-inventory policy which means that order intervals are constant (i.e. one week), and quantities ordered differ from one order to the next (Gisela Kastner, personal communication, July 04, 2003a; Kluck, 1998). Unless there is an unusual situation necessitating expedited shipping and if there is demand, the thermostats are delivered once a week to TCG Systemtechnik (Gisela Kastner, personal communication, July 04, 2003a; Markus

therefore, it is not considered in this dissertation. It seems that VW in Kassel is not a critical member of this chain because the amounts shipped to Kassel were very little compared to the ones shipped to Audi Hungaria (see Gisela Kastner, July 07, 2003). 64

TCG Systemtechnik refers to this part as “Kühlwasserregler” (part number: S005217ZK) and Wahler calls it “Thermostat W18” (part number: 4802.87/14).

65

The structure of a supply call-off is often standardized according to VDA 4905, and the structure of a precise supply call-off is often standardized according to VDA 4915 (VDA, n.d.). Both are standards defined by the Verband der Automobilindustrie. Audi uses both standards (VW Group Supply.com, 2003a). Two exemplary supply call-offs are shown in Figure 6 and Figure 7 in Appendix A.6.

108

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

Bedic, personal communication, July 04, 2003). According to Bedic and Kastner (personal communication, July 04, 2003a; personal communication, July 04, 2003a), the latter party is responsible for paying and for organizing the shipment of the thermostats. It outsources the organization and execution of the deliveries to the LSP Schenker AG which transports the parts to TCG Systemtechnik’s plant in Micheldorf (Austria) by truck (Gisela Kastner, personal communication, July 04, 2003a; Markus Bedic, personal communication, July 04, 2003). Transportation between Wahler and TCG Systemtechnik usually takes three days (one day with expedited shipping) (Gisela Kastner, personal communication, July 04, 2003a; Markus Bedic, personal communication, July 04, 2003). This can be regarded as a long lead time66, and it might explain why TCG Systemtechnik keeps a safety stock of five working days worth of thermostats (Andreas Heim, July 10, 2003) which can be considered as high: According to Müller (personal communication, June 26, 2003), a safety stock of approximately 1.5 days can be regarded as average; and it is common practice to keep a larger than usual safety stock of those parts that have a long lead time. It cannot be generally stated how many thermostats equal five working days worth of supply because this depends on Audi Hungaria’s demand for water pumps. In other words, the quantity of thermostats held as safety stock is not static, but develops dynamically over time depending on Audi Hungaria’s demand (Gisela Kastner, personal communication, August 08, 2003). TCG Systemtechnik uses the thermostats from Wahler to manufacture water pumps67 for Audi Hungaria, whereby it aims to keep a safety stock of five days worth of supply of water pumps (Andreas Heim, July 10, 2003; Gisela Kastner, personal communication, July 04, 2003a). Audi Hungaria is responsible for paying the transportation costs and for physically moving the water pumps from TCG Systemtechnik to its plant in Hungary (Gisela Kastner, personal communication, July 04, 2003a). However, Audi Hungaria does not carry out the transportation itself; it outsources this activity to the LSP VW T which buys transportation services from Schachinger Logistik Holding GmbH & Co KG (Schachinger) (Gisela Kastner, personal communication, July 04, 2003a; Katalin Szorady, personal communication, July 16, 2003). Unless there is an unusual situation which necessitates expedited shipping and if there is demand, this LSP ships the water pumps once a week, ideally in multiples of the lot size 150 to Ingolstadt by

66

Only information on transportation lead time was made available and, therefore, this serves as an approximation of total lead time. However, it has to be stressed that there might be situations in which transportation lead time is substantially shorter than total lead time; especially if the supplier does not have the ordered product in stock. Total lead time is the time between making an order and having the product available for production at the assembly line.

67

TCG Systemtechnik refers to this part as “ZB WAPU Gehäuse W 18 DG” (part number: S005210) and Audi Hungaria calls it “Wasserpumpe W18 1/V8 4.0l 4V TDI Common Rail” (part number: 057121011E).

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

109

truck where this freight is pooled with other freight bound for Audi Hungaria. The weekly transport from Ingolstadt to Györ is by train (Gisela Kastner, personal communication, July 04, 2003a; Katalin Szorady, personal communication, July 16, 2003). This intermodal transportation from Micheldorf to Györ takes four days which can be considered a long transportation lead time; especially if one bears in mind that by truck it only takes about six hours from Micheldorf to Györ (Katalin Szorady, personal communication, July 16, 2003). This long transportation lead time could explain why Audi Hungaria keeps a high safety stock of water pumps (five days worth of supply, i.e. 150 water pumps) (Katalin Szorady, personal communication, July 16, 2003). When ordering water pumps from TCG Systemtechnik, Audi Hungaria practices a (t,S)-inventory policy (Katalin Szorady, personal communication, July 16, 2003): Once a week, Audi Hungaria orders a variable amount of water pumps with a supply call-off via EDI which contains weekly demand information for the next six months. The quantities stated are not binding and can be changed with every supply call-off (Katalin Szorady, personal communication, July 16, 2003). In other words, each supply call-off supersedes its predecessor. Audi Hungaria expects that its suppliers can handle a 15 percent increase or decrease of quantities ordered from one supply calloff to the next. If quantities are reduced by more than that, Audi Hungaria has to pay for the quantities demanded; even though it does not need them at the moment. If quantities are increased by more than 15 percent, Audi Hungaria has to cover all additional costs (e.g. costs that arise because of extra shifts or expedited shipping) (Katalin Szorady, personal communication, July 16, 2003). The water pumps are used by Audi Hungaria to manufacture V8 4.0l diesel engines for Audi Neckarsulm. Since 2001, all engines for Audi cars are assembled by Audi Hungaria (Thomas Müller, personal communication, July 02, 2003); thus, also the V8 4.0l diesel engine for the A8. According to Szorady (personal communication, July 16, 2003), Audi Hungaria does not keep an inventory of engines – not even a safety stock. The engines are produced according to weekly supply call-offs which Audi Neckarsulm sends to Audi Hungaria by EDI, and the engines are delivered to Audi Neckarsulm shortly after their production (Katalin Szorady, personal communication, July 16, 2003). For transportation to Neckarsulm, the engines are loaded on frames, six at a time, by the LSP Schenker AG (Jörg Binick, personal communication, July 07, 2003). Then another LSP, DB Cargo AG, takes over and transports the engines by train on a daily basis first from Györ to Audi Ingolstadt, where the engines are separated from other freight transported on the same train, and then from Ingolstadt to Neckarsulm, its final destination (Jörg Binick, personal communication, July 07, 2003; see Figure 56). This delivery by train takes two days (Jörg Binick, personal communication, July 07, 2003). Currently, there are three trains per

110

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

day from Györ to Ingolstadt transporting Audi cars and engines for various VW Group companies (including Audi) as well as three or four trains going in the opposite direction transporting engine and vehicle components; between Ingolstadt and Neckarsulm, there are two trains per day in each direction (Jan Baumung, personal communication, July 29, 2003; Katalin Szorady, personal communication, July 16, 2003; Klaus Überrigler, personal communication, July 09, 2003; Mr. Nesch, personal communication, July 09, 2003). Audi Neckarsulm pays for regular transportation68 which is organized by Audi Hungaria (Jörg Binick, personal communication, July 07, 2003). Audi Hungaria outsources the organization to VW T which purchases transportation services from DB Cargo AG (Andrea Maggio, personal communication, July 10, 2003; Jörg Binick, personal communication, July 07, 2003). In order to provide rail transportation between Germany and Hungary, DB Cargo AG joins forces with the Austrian rail company ÖBB and the Hungarian rail companies GYSEV and MAV (DB Cargo, n.d.). Since transportation lead time is quite long (two days), Audi Neckarsulm keeps a high safety stock of engines: three days worth of supply, i.e. ninety units (Jörg Binick, personal communication, July 07, 2003).69 Keeping this amount is quite expensive: If this amount is held, just the cost of tied-up capital amounts to EUR 55,687.50 per year.70 Like the other members of this supply chain, Audi Neckarsulm pursues a (t,S)-inventory policy, i.e. it orders variable amounts of engines in fixed intervals.71 The maximum amount of A8s with a V8 diesel engine that can be produced per week is 150, i.e. thirty can be produced five days a week. The maximum daily amount of A8s with a V8 diesel engine is sixty; however, there is the restriction that only every other A8 can get a V8 diesel engine (Jörg Binick, personal communication, July 07, 2003). In the monitored period of time, the average weekly demand for V8 diesel engines (and, thus, probably also for A8s with this engine) was lower than the maximum possible amount: It was only 93.29 [calculated on data extracted from BETA93 (list LUA4000A) on August 01, 2003] which is 41.72 percent less than Audi Hungaria’s average weekly water pump demand of 160.06 pieces (calculated on data provided by Katalin Szorady, personal

68

If expedited shipping is necessary, the party which caused the problem has to pay (Jörg Binick, personal communication, July 07, 2003).

69

Both Audi Hungaria and Audi Neckarsulm list their safety stock not only in terms of days of supply given a certain demand, but also in terms of a certain quantity. This would make sense if demand of these two companies was stable in the monitored period of time, but as shown in Table 18 their demand varies. Consequently, the amount which represents a certain number of days of supply also varies.

70

For calculating this amount, see Appendix A.5.

71

Binick (personal communication, July 07, 2003) claims that Audi Neckarsulm pursues a (s,S)-inventory policy; however, this seems to be a mistake because Audi Neckarsulm orders weekly, i.e. in fixed intervals by supply calloff which is sent to Audi Hungaria.

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

111

communication, August 13, 2003) in the same period of time. This indicates that orders received by Audi Hungaria for its end product (engines), and its orders to TCG Systemtechnik for parts necessary to make that end product (water pumps) are out of tune which might be a sign that the bullwhip effect is present in this supply chain (see Section 4.3.1). This ends the description of the supply chain of the V8 4.0l diesel engine, or rather its critical path. A description of how a critical path is determined in general and in particular follows along with a description of how the critical path of the engine supply was selected. In the GEKO project, a SCM project which Audi is currently carrying out and which pilots the ICON-SCC72, the following criteria were defined in order to determine the critical path and, thus, the companies participating in the project (Scheidler, 2003a): •

high value of a part or module,

•

long production and transportation time,

•

past bottlenecks and special measures,

•

deliveries can only be guaranteed by holding large amounts of inventory,

•

high value creation early in the supply chain,

•

versioning early in the supply chain,

•

large demand variability,

•

increase in demand and/or a bottleneck can be expected for the future,

•

frequent engineering changes, and

•

versions that are rarely in demand.

In order to determine the critical path of a supply chain, these items are not to be applied as strict rules; but they are to serve as a guide, and the intuition and experience of the specifically involved people can also influence the decision. When selecting Audi’s engine supply chain, or more specifically certain parts of it for the GEKO project, these criteria played only an ancillary role; however, they are listed so as to give an impression of how critical paths could be selected in general. The main reason why the critical path of the engine supply chain described in this section has been selected for the GEKO project was that the agency BWL research & consulting,

72

GEKO stands for “Ganzheitliche Ertragsorientierte Kettenoptimierung” and is one of several SCM projects that Audi currently pursues. The project is divided into two parts: One part includes product and value analyses and, thus, focuses on purchasing issues. The other part deals primarily with logistics, more precisely with SCMo (Scheidler, 2003b).

112

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

which conducted an analysis of this chain or rather the selected path for Audi, asked for this supply chain to be included in the project (Roland Scheidler, personal communication, July 11, 2003). It can be assumed that there must have been problems along the critical path of the engine supply chain as shown in Figure 33; otherwise, Audi would not have bought the services of BWL research & consulting to analyze this path. Two problems were mentioned to the author. Heim of TCG Systemtechnik (personal communication, July 08, 2003) said that his company wanted to include Wahler because Wahler had delivery problems (see also Gisela Kastner, personal communication, July 04, 2003b). Bedic of Wahler (personal communication, July 04, 2003) said that these delivery problems were not caused by Wahler, but by one of its suppliers which was not able to deliver the right quantity at the right time in the right quality. The second problem, a bottleneck, was due to delivery problems of one of Audi Hungaria’s suppliers: Per week, a maximum amount of only 150 turbo chargers could be delivered to Audi Hungaria and, consequently, the maximum number of Audi A8s with a V8 diesel engine was restricted to that amount per week.73 Originally, 250 units were planned (Jörg Binick, personal communication, July 07, 2003; Katalin Szorady, personal communication, July 16, 2003). These two problems and possibly others that the author has not been made aware of might have adversely affected and might still affect a number of companies in the engine supply chain. For example, due to these problems, special measures like expedited shipments and extra shifts were necessary and might still be (Thomas Hanin, personal communication, July 10, 2003). Wahler, for instance, had to pay EUR 700.80 for expedited shipments in the first half of 2003; however, it passed these costs on to the supplier which caused the problem (Markus Bedic, personal communication, July 17, 2003). Audi Neckarsulm and Audi Hungaria claim that they did not incur any additional expenses due to the problems and the resulting backlogs (Jörg Binick, personal communication, July 07, 2003; Katalin Szorady, personal communication, July 16, 2003). TCG Systemtechnik is not able to make a statement on the amount of backlog costs it had to bear because this company does not record these costs on a per customer basis (Gisela Kastner, personal communication, July 04, 2003a). Since BWL research & consulting found out that special measures were taken (Thomas Hanin, personal communication, July 10, 2003), there must have been extra costs. The apparent

73

The turbo charger bottleneck has been resolved by now. Nonetheless, maximum production has not been increased to the originally planned 250 units of V8 4.0l diesel engines because there was not enough demand. It is even debated to reduce the maximum weekly production to 100 instead of 150 pieces (Katalin Szorady, personal communication, July 16, 2003).

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

113

contradiction that some supply chain members claim not to have had extra costs due to backlogs could be explained with the practice that costs are always passed on to those suppliers which caused the problem, and that those suppliers are not part of the analyzed path. According to the members of this supply chain, passing on costs seems to be common practice. For example, Binick of Audi Neckarsulm (personal communication, July 07, 2003) alleged that Audi Hungaria has to bear all backlog costs because it committed itself to delivering certain amounts per week. Binick said that if Audi Hungaria is not able to deliver, and if this problem is caused by one of its suppliers, then this supplier probably has to cover these costs. Szorady (personal communication, July 16, 2003) supported this statement by claiming that Audi Hungaria did not incur any additional expenses, but she also stated that there were expedited shipments; especially right after series production started. Therefore, it can be assumed that these expenses have been passed on to other companies as well. Kastner (personal communication, July 04, 2003) said that TCG Systemtechnik covers only the costs for problems that it caused; all other costs are passed on to suppliers like Wahler. Backlogs can be regarded as an indicator of problems in the supply chain. Unfortunately, no member in this chain, except Audi Neckarsulm, provided data on backlog or reorder quantities over the monitored period of time. The backlogs which Audi Neckarsulm states in its supply calloffs to Audi Hungaria between week 10 and week 26 of 2003 are shown in Figure 34.

160 140 quantity

120 100 80 60 40 20 week 26

week 25

week 24

week 23

week 22

week 21

week 20

week 19

week 18

week 17

week 16

week 15

week 14

week 13

week 12

week 11

week 10

0

time Figure 34: Backlogs Quoted in Audi Neckarsulm’s Supply Call-offs to Audi Hungaria between Week 10 and Week 26 of 2003 (Source: Based on data extracted from BETA 93 (list LBE9080A) on July 30, 2003)

Except for week 11, there were backlogs at Audi Neckarsulm in every week (see Figure 34) which might be an explanation for the low inventory level of engines at Audi Neckarsulm (see Section 4.2.2). This corresponds with the results of the analysis carried out by BWL research &

114

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

consulting: The agency found that a large number of engines was transported from Györ to Neckarsulm with expedited shipping (Thomas Hanin, personal communication, July 10, 2003) – this special measure was probably necessary because Audi Hungaria was running late on its deliveries; and this way, transportation could be speeded up. The reasons for this situation can be manifold. One reason could be delivery problems further upstream the supply chain. However, the available information is not sufficient for determining the causes of the backlogs at Audi Neckarsulm. Irrespective of what caused the backlogs, the situation at Audi Neckarsulm can be regarded as an indicator of problems upstream the supply chain. For example, if Wahler is late with its delivery of thermostats to TCG Systemtechnik, TCG Systemtechnik might be late with its deliveries to Audi Hungaria, and Audi Hungaria might, therefore, be late with its deliveries to Audi Neckarsulm.

4.2.2 Analysis of Inventory Levels in the Supply Chain of the Audi A8 V8 4.0l Diesel Engine This section analyzes how the inventory levels of thermostats, water pumps, and engines developed in the engine supply chain between week 10 and week 26 of 2003. Inventory levels are looked at in detail because excessive or insufficient inventory can be a consequence of the bullwhip effect (see Section 2.2.4.2) and, thus, also be an indicator of the fact that a supply chain suffers from this effect. According to Bedic (personal communication, July 04, 2003), Wahler only produces whatever TCG Systemtechnik demands and does not even keep any safety stock of thermostats. The analysis of the inventory data on thermostats provided by Wahler (Markus Bedic, personal communication, July 17, 2003) results in a different picture: It indicates that the average inventory level between week 10 and week 26 of 2003 is 221.42 units74 which is approximately equal to the maximal output of 4.4 days of production.75 This average inventory level can be regarded as high because Wahler claims to only produce thermostats based on actual demand by TCG Systemtechnik and not to carry any inventory (not even a safety stock) (Markus Bedic, personal communication, July 17, 2003). The cost of capital tied-up in the average inventory level is EUR 298.92 per year.76 This is a rather small amount and, therefore, it seems to be negligible; however, one should not forget two things. First, one reason why the cost of capital tied-up is so low is

74

This number has been calculated based on data received from Wahler (Markus Bedic, personal communication, July 17, 2003). All figures were added up and divided by the total number of figures available (see Table 5, Appendix A.7).

75

Wahler has a maximum production capacity of fifty thermostats per day, i.e. 250 per week (ICON, 2003d).

76

For calculating this amount, see Appendix A.5.

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

115

that thermostats are relatively low-cost parts (EUR 15.00 per piece)77 compared to other parts (see Table 17). Second, it should be considered that the cost of capital tied-up is only one part of the total inventory carrying costs which is the sum of the costs for storage, obsolescence, insurance, taxes, and capital tied-up (Bowersox & Closs, 1996). Consequently, total expenses incurred due to carrying that amount of inventory are probably much higher; however, Wahler did not provide complete information on its total inventory carrying costs. Cost of capital tied-up and possibly other components of total inventory carrying costs could be reduced if the amount of inventory kept was lowered. The development of Wahler’s inventory level of thermostats in the monitored period of time is visualized in Figure 35.

700

quantity

600 500 400 300 200

06-23-2003

06-16-2003

06-09-2003

06-02-2003

05-26-2003

05-19-2003

05-12-2003

05-05-2003

04-28-2003

04-21-2003

04-14-2003

04-07-2003

03-31-2003

03-24-2003

03-17-2003

03-10-2003

03-03-2003

100 0

time Figure 35: Wahler’s Inventory Level of Thermostats between Week 10 and Week 26 of 2003 (Source: Based on Markus Bedic, personal communication, July 17, 2003)

The above figure also visualizes that Wahler’s inventory level is continuously high between April and June 2003, i.e. after the bottleneck had been resolved which prevented Wahler from delivering to TCG Systemtechnik (Gisela Kastner, personal communication, August 08, 2003). The development of TCG Systemtechnik’s inventory level of thermostats in the monitored period of time is shown in Figure 36. In May and June 2003 there is a trend towards an increasing inventory level. Three times in the monitored period of time (between March 13th and March 24th, between April 04th and April 24th, and between May 13th and May 27th), TCG Systemtechnik did not have any thermostats in stock.

77

According to Andreas Heim, personal communication, July 10, 2003.

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

06-23-2003

06-16-2003

06-09-2003

06-02-2003

05-26-2003

05-19-2003

05-12-2003

05-05-2003

04-28-2003

04-21-2003

04-14-2003

04-07-2003

03-31-2003

03-24-2003

03-17-2003

03-10-2003

900 800 700 600 500 400 300 200 100 0 03-03-2003

quantity

116

time Figure 36: TCG Systemtechnik's Inventory Level of Thermostats between Week 10 and Week 26 of 2003 (Source: Based on Gisela Kastner, personal communication, July 07, 2003)

TCG Systemtechnik’s average inventory level of kept thermostats between week 10 and week 26 of 2003 is very similar to Wahler’s: It is 226.86 units.78 Whether this amount is to be interpreted as high, low, or just right depends on Audi Hungaria’s demand for water pumps. Audi Hungaria’s average weekly demand in the monitored period of time was 160.06 water pumps (calculation based on Katalin Szorady, personal communication, August 13, 2003). Hence, 226.86 water pumps are on average 7.01 days of supply which can be regarded as an excessive average amount of inventory. This is backed up by the Kastner’s statement that TCG Systemtechnik had a high inventory level of thermostats for at least some time in the monitored period (personal communication, August 04, 2003a; personal communication, August 08, 2003). The cost of capital tied-up for carrying this average amount of inventory is EUR 306.26 per year79; however, this figure has to be interpreted the same way as Wahler’s cost of capital tied-up in thermostats. The fact that the inventory level of thermostats at both Wahler and TCG Systemtechnik is high at times and low at others could be interpreted as circumstantial evidence for the existence of the bullwhip effect in this supply chain (see Section 2.2.4.2). Furthermore, the increase in inventory levels over time could be explained by the fact that the production of the V8 diesel engine only started in March 2003 and that the negative effects of the bullwhip effect (e.g. excessive inventory levels) only developed over time. However, it has to be stressed that answering the question of why inventory levels are high at times and low at others with the existence of the bullwhip effect

78

This number was calculated based on data received from TCG Systemtechnik (Kastner, personal communication, July 07, 2003). All figures were added up and divided by the total number of figures available (see Table 5, Appendix A.7).

79

For the calculation of this amount, see Appendix A.5.

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

117

is only based on mere presumption – the existence of the bullwhip effect in the engine supply chain still has to be proven in Section 4.3.1. Figure 37 shows how TCG Systemtechnik’s inventory level of water pumps developed between week 10 and week 26 of 2003.

350 300 quantity

250 200 150 100 50 06-23-2003

06-16-2003

06-09-2003

06-02-2003

05-26-2003

05-19-2003

05-12-2003

05-05-2003

04-28-2003

04-21-2003

04-14-2003

04-07-2003

03-31-2003

03-24-2003

03-17-2003

03-10-2003

03-03-2003

0

time Figure 37: TCG Systemtechnik’s Inventory Level of Water Pumps between Week 10 and Week 26 of 2003 (Source: Based on Gisela Kastner, personal communication, August 04, 2003b)

What is striking about Figure 37 is that the inventory level seems to oscillate between zero and 150 or 300 most of the time, when 150 is Audi Hungaria’s lot size (Martin Hackner, personal communication, July 28, 2003). There are two possible interpretations for the way TCG Systemtechnik’s inventory level of water pumps developed over time. First, this company might carry a high inventory of 150 or 300 units for the time span between the completion of production and delivery, i.e. only for a short period of time. This interpretation is likely to apply if the high amount of inventory is held only for a short period of time like, for example, between May 09 and May 11, 2003 or between May 14 and May 15, 2003. Second, it could be that Audi Hungaria had announced the need of a certain amount at a certain future point in time, and then this demand was reduced in the final supply call-off, or in one of the last supply call-offs, and so TCG Systemtechnik suddenly might have had surplus water pumps.80 This interpretation is also supported by Kastner’s statement (personal communication, August 04, 2003; personal communication, August 08, 2003) that Audi reduced its demand several times for a particular week in the final supply call-off, or in one of the last supply call-offs; and that this caused high

80

Only if Audi Hungaria decreases its demand by more than 15 percent, it is obliged to take the water pumps even though it does not need them in that moment (Katalin Szorady, personal communication, July 16, 2003).

118

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

inventory in the chain, for instance in June 2003. This interpretation is likely to apply if TCG Systemtechnik holds the high amount of inventory for a longer period of time which was, for example, the case between May 23 and June 04, 2003. TCG Systemtechnik’s average inventory level is 73.10 water pumps81 which is 2.28 days of supply if Audi Hungaria’s average weekly demand of 160.06 water pumps (calculated based on Katalin Szorady, personal communication, August 13, 2003) is assumed. This means that TCG Systemtechnik was not able to meet its safety stock requirement of five days of supply for most of the time in the monitored period. Hence, it can be concluded that TCG Systemtechnik’s inventory level of water pumps was low on average between week 10 and week 26 in 2003. This company’s inventory level of water pumps was higher than Audi Hungaria’s average weekly demand only on some days (between April 15 and April 17, between May 17 and May 19, and on June 3) and can, therefore, be regarded as relatively high. Due to a lack of data, the development of Audi Hungaria’s water pump inventory between March and June 2003 cannot be described. According to Szorady (personal communication, July 16, 2003), inventory data is not recorded by Audi Hungaria which, in the opinion of the author, is rather unlikely to be actually the case. The reluctance to supply these figures might have something to do with the apparently overdimensioned order amounts of Audi Hungaria. However, this is just a subjective assumption of the author based on observations and personal impressions while conducting the interviews. All that can be calculated for this company is the annual cost of capital tied-up in the safety stock of 150 water pumps: It is EUR 496.6782 whereby this figure has to be interpreted the same way as cost of capital tied-up in thermostats and water pumps at Wahler and TCG Systemtechnik. Additionally, it has to be considered that this is only the amount of capital tied-up in safety stocks – the assumption does not seem too far-fetched that in addition to these assets of the last resort there are additional stocks. Therefore, it can be presumed that total inventory carrying costs are probably much higher. The current safety stock of water pumps held at Audi Hungaria could possibly be reduced if transportation lead time was shortened (Carter & Ferrin, 1995). A transportation lead time reduction to one day (or, more precisely, six hours) seems obtainable if TCG Systemtechnik delivered directly to Audi Hungaria. This would make sense because TCG Systemtechnik is located half way between Audi Hungaria and Audi Ingolstadt; as a matter of fact, making this

81

This number has been calculated based on data received from TCG Systemtechnik (Kastner, personal communication, July 07, 2003). All figures were added up and divided by the total number of figures available (see Table 5, Appendix A.7).

82

For calculating this amount, see Appendix A.5.

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

119

change is currently in debate (Katalin Szorady, personal communication, July 16, 2003). If transportation lead time were shortened to one day, it does not seem too bold to reduce the safety stock to two days of supply.83 Thus, the annual cost of capital tied-up in this safety stock would only be EUR 198.67, i.e. it would be reduced by EUR 298.00 or about forty percent. In monetary terms, this reduction does not seem world-shattering at all; however, percentage-wise, the reduction is substantial. Furthermore, it has to be kept in mind that the cost of capital tied-up only serves as an indicator of total inventory carrying costs. These total costs are probably much higher than the cost of capital tied-up; and if one could cut total costs, this would probably make a difference. A reduction of total inventory carrying costs by forty percent seems unrealistic; but depending on the extent of variable cost components, the decrease might be substantial. Audi Hungaria claims not to store any engines but to deliver them right after production is finished. Three days of supply of engines (i.e. 90 units) are stored at Audi Neckarsulm as a safety buffer. However, as shown Figure 38 the inventory level was below this mark for most of the

06-22-2003

06-15-2003

06-08-2003

06-01-2003

05-25-2003

05-18-2003

05-11-2003

05-04-2003

04-27-2003

04-20-2003

04-13-2003

04-06-2003

03-30-2003

03-23-2003

03-16-2003

03-09-2003

200 180 160 140 120 100 80 60 40 20 0 03-02-2003

quantity

time.

time Figure 38: Audi Neckarsulm’s Inventory Level of V8 4.0l Diesel Engines between Week 10 and Week 26 of 2003 (Source: Based on data extracted from BETA93, list LUA4000A, on July 30, 2003 and August 4, 2003)84

83

The reasoning behind this statement is the following: Currently, the members of this supply chain maintain a safety stock level (expressed in days of supply) which exceeds lead time by at least one day. This would still be the case if the suggested reduction of lead time and of safety stock was realized.

84

BETA93 is an IT-system for archiving data (Hinsenkamp, 2002a). In this case, the data extracted from BETA93 was originally from a system called “Materialfluss-Bewegungsverarbeitungs- und Bestandsführungs-System” (MABES), which is an inventory management system (Hinsenkamp, 2002b).

120

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

The average amount of engines held between March and June 2003 was 82.6785 which amounts to EUR 51,152.0686 in cost of capital tied-up per year. At first glance, this might seem alarming because this is less than the desired safety stock level of ninety engines. However, it can be explained with the fact that the engine production in series was just started in March 2003 and, therefore, a safety stock needed to be built up first. As Figure 38 shows, there is an overall trend towards an increasing inventory level. However, in June 2003 inventory was low again – even below 90 units – for a while before it increased again. Table 17 summarizes key inventory figures that have been discussed in this section. It includes the annual cost of capital tied-up in the average inventory level, the product unit price, the interest rate, the average inventory level as well as the minimum and the maximum inventory level. Table 17 shows several things. First, it can be seen that the cost of capital tied-up in inventory is the highest at Audi Neckarsulm; although its average inventory level is comparably small. The reason for this is that engines are more expensive than water pumps and thermostats. This demonstrates that from the perspective of the entire supply chain, holding large inventories upstream in the supply chain where parts are less expensive might sometimes be the better choice; although this would harm those suppliers that have to carry the high inventory.87 Second, there seems to be the tendency that the average inventory held upstream in the supply stream is larger than the average inventory held downstream. However, it can be assumed that this is not the result of supply chain optimization efforts, but that this rather reflects that suppliers get stuck with a high inventory of parts and/or of end products if their customers suddenly reduce order quantities. This happened several times in the engine supply chain; especially in June 2003 (Gisela Kastner, personal communication, August 04, 2003; Gisela Kastner, personal communication, August 08, 2003). The reason why Wahler’s average inventory level of thermostats is smaller than TCG Systemtechnik’s might be that the first four weeks of TCG Systemtechnik’s supply call-offs to Wahler are binding. This means that in a current week, TCG Systemtechnik has to take the amount of thermostats it stated in the supply call-off four weeks ago even if Audi Hungaria decreases its demand and if TCG Systemtechnik, therefore, gets stuck with surplus supplies of thermostats and/or water pumps (Gisela Kastner, personal communication, July 04, 2003a).

85

This number has been calculated based on data received from Audi Neckarsulm [data extracted from BETA93 (list LUA4000A) on July 30, 2003 and August 4, 2003]. All figures were added up and divided by the total number of figures available (see Table 5, Appendix A.7).

86

For the calculation of this amount, see Appendix A.5.

87

The issue of global versus local optimization is often discussed within the framework of game theory. For more on this, see Section 4.3.3 and Appendix A.2.

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

121

Table 17: Summary of Key Inventory Figures88 Audi Neckarsulm

Audi Hungaria

TCG Systemtechnik

Wahler

Cost of capital tied-up (p.a.) at the average inventory level

EUR 51,152.06 (engines)

Not available

EUR 306.26 (thermostats)

EUR 298.92 (thermostats)

Product unit price

EUR 6,875.00 (engines)

EUR 243.23 (water pumps) EUR 6,875.00 (engines)

EUR 15.00 (thermostats)

EUR 36.97 (water pumps)

EUR 36.97 (water pumps)

EUR 15.00 (thermostats)

Interest rate

9 percent

9 percent

9 percent

9 percent

Average inventory level

82.67 (engines)

Not available

226.86 (thermostats)

221.40 (thermostats)

73.10 (water pumps) Minimum inventory level

52 (engines)

Not available

0 (thermostats)

0 (thermostats)

0 (water pumps) Maximum inventory level

182 (engines)

Not available

788 (thermostats)

650 (thermostats)

300 (water pumps)

Third, the range between minimum and maximum inventory level also increases further upstream in the supply chain as shown in Table 17. The described development of the inventory levels of thermostats, water pumps, and engines in the engine supply chain is closely related to the orders which the members of this supply chain placed in the monitored period of time. These orders will be the focus of the next section.

4.2.3 Analysis of Orders Placed by the Companies in the Supply Chain of the Audi A8 V8 4.0l Diesel Engine In this section, the orders placed by the companies in the engine supply chain between week 10 and week 26 of 2003 will be analyzed because this data is the basis for calculating the bullwhip effect. The order quantity represents a company’s net demand.

88

Note: The interest rate of nine percent is the one which Audi Neckarsulm uses. The other companies might use different rates, for example, Wahler uses five percent (Markus Bedic, personal communication, July 30, 2003). However, when calculating the cost of capital tied up in inventory, the same rate was used for all companies to ensure comparability.

122

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

The supply call-offs which the members of this supply chain send to their suppliers would be ideal for calculating the bullwhip effect because these documents constitute orders in this supply chain and, thus, reflect net demand. However, only Audi Neckarsulm provided this information. In order to guarantee comparability of the data, the amounts each company received from its supplier were used as an approximation of net demand which is expressed in the supply call-off – this was the only information that all companies provided. In the following paragraphs, the received information will be described more closely. As an approximation of the net demand of Audi Neckarsulm, the amounts which arrived at the goods receiving of this company are used. This information was extracted from BETA93 (list LUA4000A) on August 01, 2003 (originally, it was stored in MABES). The data used is only an approximation of net demand for two reasons. 1. Engines were sometimes sent back to Hungary because they did not meet the defined quality standards (see BETA93-list LUA4000A). These engines are included twice in list LUA4000A: Once when they arrived in a defective state, and a second time when they were delivered in the correct quality. In other words, the sum of engines received is too large. According to the data shown in the BETA93-list LUA4000A, only 35 units were sent back in the analyzed period of time and 1,586 were received in total. The deliveries received per week shown in Table 18 could not be adjusted by that amount because the BETA93-list LUA4000A does not show when an engine was delivered a second time – it only shows when an engine was sent back. However, since only a very small percentage of received engines was sent back, this flaw of the used data can be regarded as negligible. 2. Audi Hungaria was not capable of always delivering the amount which had been ordered, as is shown in Figure 34. That means that the sum of engines received by Audi Neckarsulm is too small compared to the quantity ordered. However, the supply call-offs for the two variants of the V8 4.0l diesel engine from June 23, 2003 [extracted from BETA 93 (list LBE9080A) on July 30, 2003]89 indicate that by the end of June 2003 there was only a backlog of 90 engines. Since this amount is small compared to the delivered amount, this flaw of the data can be regarded as negligible as well. Consequently, one can conclude that the quantity of engines which arrived at Audi Neckarsulm’s goods receiving each week is an adequate approximation of Audi Neckarsulm’s weekly net demand stated in the supply call-offs.

89

This data originally came from a system called “Lieferabrufverteilungs- und Feinsteuerungssystem (LAFES), whose purpose it is to create and update supply call-offs (Hinsenkamp, 2003c).

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

123

Audi Hungaria provided information on the quantities of water pumps that it received from TCG Systemtechnik (see Katalin Szorady, personal communication, August 13, 2003). The information from Audi Hungaria is also mirrored in information from TCG Systemtechnik on the amounts of water pumps it delivered to Audi Hungaria (see Gisela Kastner, personal communication, July 07, 2003): In the two items of information, the total amount delivered is the same, but the distribution of the deliveries over the monitored weeks differs. This is normal because TCG Systemtechnik might have shipped out water pumps in one week, and Audi Hungaria might have received these parts in the following week. If the required quantities are equal or at least close to the received quantities, then the received quantities are a good approximation of Audi Hungaria’s net demand. However, there is the possibility that the received and ordered amounts are not the same. If there is a difference between these two quantities, the amount ordered probably exceeds the amount received because of backlogs. As a result, the bullwhip effect would differ from the one calculated based on the available data. TCG Systemtechnik provided information on the amount of thermostats it received from Wahler, or more specifically on its increase of thermostat stocks (see Gisela Kastner, personal communication, July 07, 2003).90 This amount is only an approximation of TCG Systemtechnik’s order quantities and, thus, of its net demand: the net demand might actually be larger than the amounts received due to backlogs. If this is the case, the bullwhip effect would differ from the one calculated based on the available data. For the rest of this dissertation, the data which is available for calculating the bullwhip effect will be referred to as order quantities; although these quantities are only an approximation thereof. Table 18 gives an overview of the weekly order quantities by Audi Neckarsulm, Audi Hungaria, and TCG Systemtechnik. Wahler is not included because the orders which this company places to its suppliers are not part of the critical path. The table shows weekly order quantities because the members of this supply chain send out supply call-offs at weekly intervals, and supply calloffs constitute orders in this supply chain. A more detailed overview of orders (or more specifically deliveries) is shown in Appendix A.7. This table shows deliveries received by the members of this supply chain on a daily instead of a weekly basis.

90

Wahler also provided information on the quantities it delivered to TCG Systemtechnik (Markus Bedic, personal communication, July 17, 2003). The amount that Wahler claims to have delivered between March and June of 2003 is smaller than the amount that TCG Systemtechnik claims to have received in the same period of time: Wahler says it delivered 3,819 thermostats (Markus Bedic, personal communication, July 17, 2003), and TCG Systemtechnik says it received 3,934 water pumps (Gisela Kastner, personal communication, July 07, 2003). Although this difference ought to be explained, this point has not been investigated any further.

124

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

Table 18: Orders of Audi Neckarsulm, Audi Hungaria, and TCG Systemtechnik between Week 10 and Week 26 of 2003 (Weekly Amounts) k = 1 (OEM)

k = 2 (tier 1)

k = 3 (tier 2)

Audi Neckarsulm

Audi Hungaria

TCG Systemtechnik

V8 4.0l diesel engines

water pump W18

thermostat

1

week 10

16

47

372

2

week 11

9

0

0

3

week 12

8

0

0

4

week 13

33

150

218

5

week 14

53

150

296

6

week 15

36

285

0

7

week 16

64

0

292

8

week 17

130

300

277

9

week 18

84

0

192

10

week 19

164

449

0

11

week 20

129

300

441

12

week 21

136

300

294

13

week 22

130

150

384

14

week 23

140

150

188

15

week 24

154

150

384

16

week 25

102

140

384

17

week 26

198

150

SUM

1,586

SUM

2,721 SUM

212 3,934

MEAN

93.294

MEAN

160,059 MEAN

231.412

STD. DEV.

58.178

STD. DEV.

126.912 STD. DEV.

146.645

VARIANCE

3,384.678

VARIANCE

16,124.408 VARIANCE

21,504.713

(Source: Based on data extracted from BETA93 (list LUA4000A) on August 01, 2003; Gisela Kastner, personal communication, July 07, 2003; Katalin Szorady, personal communication, August 13, 2003)

As can be seen in Table 18, total demand and mean demand increase upstream in the engine supply chain. This is in need of explanation. The first explanation as to why the demand for thermostats and water pumps is larger than the demand for engines might be that the water pumps (and, thus, the thermostats) are not only delivered to Audi Hungaria but also sent to VW’s location in Kassel. However, the amounts sent to this destination are very small: only 51 units between week 10 and week 26 of 2003 compared to 2,721 units to Audi Hungaria in the same period of time (Gisela Kastner, personal communication, July 7, 2003). Therefore, this can hardly serve as the sole explanation for the increase in total and mean demand upstream in the supply chain.

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

125

A second explanation might be that the water pumps (and, thus, also the thermostats) are also used for a new kind of engine which is not yet produced in series.91 However, between March and July 2003 only five of these new engines were projected to be required by Audi Neckarsulm (information extracted from BESI2 on July 26, 2003)92 and, therefore, this is hardly an explanation for the increase in total and mean demand upstream in the supply chain. A third explanation might be that to a large degree the water pumps are subject to defects or become defective in the production process of the engine. This would explain why Audi Hungaria would need to order more than one water pump per engine. However, this is most likely not to be the case since the scrap rate of water pumps is “very little” (Gisela Kastner, personal communication, August 8, 2003). A fourth explanation might be that Audi Neckarsulm announced high demand for a certain week in a supply call-off, but then it reduced demand in one of the last supply call-offs or even in the final one – at a time when the suppliers further upstream already prepared for the large demand by ordering large amounts. For example, if Audi Neckarsulm announces in the supply call-off of week x that it will need 90 engines in week x+t93 from Audi Hungaria, Audi Hungaria will need (at least) 90 water pumps from TCG Systemtechnik. Due to this demand, TCG Systemtechnik will order at least 90 thermostats from Wahler. If Audi Neckarsulm reduces its demand for engines to 20 somewhere between x and x+t, Audi Hungaria might have already ordered 90 or more water pumps by that time, thus, more than it needs to build 20 engines. This can be applied to TCG Systemtechnik’s situation as well. In such a situation, neither Audi Hungaria nor TCG Systemtechnik have an immediate use for the extra water pumps or thermostats – the surplus is most likely stored in a warehouse until needed. According to Kastner (personal communication, August 04, 2003a; personal communication, August 08, 2003), a drop in demand at Audi94 occurred in the surveyed period of time, e.g. in June 2003. This had the consequence that TCG Systemtechnik ordered too many thermostats and/or produced too many water pumps which had to be stored in the company’s warehouse until needed. 91

This new engine is the engine referred to as “Motorsorte 8451 W18 Links- und Rechtslenker” (part number: 057100015D). It was produced in series starting in week 34 of 2003 (information extracted from BESI2, June 26, 2003) and it superseded the previous two types of V8 diesel engines.

92

BESI2 is a system for determining future gross demand of parts based on the planned production of cars. In BESI2, gross demand is computed on a weekly basis (Hinsenkamp, 2002a).

93

x is a random point in time. x+t is a point in time that is t periods later than x.

94

It is unclear if Kastner refers to Audi Neckarsulm, to Audi Hungaria, or to both. She used the word “Audi” to refer to the company which frequently adjusted its demand. Figure 9 through Figure 25, Appendix A.8, show that Audi Neckarsulm frequently adjusts its demand for a particular week. Therefore, it can be assumed that Audi Hungaria does the same – at least it has the opportunity to do this because the only demand data in its supply call-offs that is finalized is the data for the week in which the supply call-off was issued.

126

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

Figure 39 visualizes the situation in which Audi Neckarsulm’s demand in a particular week is different in the final supply call-off than the demand in the same week of the supply call-off four weeks earlier.

350% 300% ratio

250% 200% 150% 100% 50% week 26

week 24

week 22

week 20

week 18

week 16

week 14

week 12

week 10

0%

time Figure 39: Ratio of Audi Neckarsulm’s Demand in a Particular Week in the Final Supply Call-off, and Demand in the Same Week in the Supply Call-off Four Weeks Earlier (Calculated between Week 10 and Week 26 of 2003) (Source: Based on data extracted from BETA 93 (list LBE9080A) on July 30, 2003)95

If the bar is below the 100 percent mark (the thicker horizontal line), it indicates that Audi Neckarsulm’s demand for a particular demand in the final supply call-off is smaller than what was stated for that week four weeks earlier in the respective supply call-off. For example, demand in week 25 should be fixed in week 21 because this is when the production schedule is theoretically fixed (Wagner, 2001). If demand for week 25 as stated in week 25 is equal to demand for week 25 as stated in week 21, the ratio would be 100 percent. However, the ratio for week 25 is 80 percent which means that demand in week 25 computed in week 25 is smaller than demand for week 25 that had been calculated four weeks earlier in week 21. The difference of demand for week 25

95

The diagram is based on an examination of the supply call-offs for one of the two diesel engines (for the one with the part number 570 100 015 B) which are jointly referred to as V8 4.0l diesel engine in this dissertation. This analysis of the supply call-offs for the engine with the mentioned part number should be representative for both types of engines because the majority of demanded engines has the mentioned part number. For a description of how the ratios were calculated, see Appendix A.8. Also shown in the appendix is the development of demand in a particular week as forecasted in the different supply call-offs (see Figure 9 through Figure 25, Appendix A.8). In order to calculate weekly demand, daily demand for each week given in the final supply call-off was added up, and the immediate demand was excluded. For weeks 10 and 22, the ratio could not be calculated due to missing data. For two reasons, the significance of these ratios is only limited. First of all, series production of this engine just started in week 10 (i.e. in the beginning of March of 2003) and therefore, demand and final production plans are probably relatively unpredictable, which can be an explanation for the frequent demand changes (Figure 9 through Figure 25, Appendix A.8,). Secondly, demand quantities, particularly in the beginning, are small, e.g. only six or twelve engines per week, and therefore, a small change of the absolute numbers will cause a drastic percentage change.

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

127

calculated in week 25 and week 21 is 20 percent which is quite a slump and might have had a negative effect on the supply chain in terms of high inventory levels. As has been noted, the weekly production schedule is ideally fixed four weeks in advance (Wagner, 2001) which is why demand for a particular week stated in the final supply call-off has been compared with the demand stated in the week of the supply call-off four weeks earlier. In reality, however, everything can change from supply call-off to supply call-off and the only quantities that are fixed are the ones stated in a week’s final supply call-off by Audi Neckarsulm (Mario Wagner, personal communication, August 01, 2003). That changes by Audi Neckarsulm are the rule rather than the exception is clearly shown in Figure 39: There was no change at only three occasions. Audi Neckarsulm’s final demand was six times lower than the previously announced demand, and it was larger six times. In defense of Audi Neckarsulm, it needs to be stated that it might be difficult to properly calculate future demand for the V8 diesel engine because this is a new part and, therefore, it is not possible to resort to historic data and to extrapolate it to the future in order to determine future demand. The demand uncertainty resulting from Audi Neckarsulm’s (and, thus, probably also Audi Hungaria’s) demand changes seems to be a serious problem in the engine supply chain. A fourth explanation for the increase of total demand and of mean demand might be that the companies in this chain (at least sometimes) order an amount that exceeds their net demand as a protection against sudden demand increases by their customers: If they have enough products in stock, they can satisfy increased demand on short notice. Or if they have enough supplies of parts, they can manufacture their final product in a small amount of time and also increase the delivered quantity on short notice. This explanation seems quite reasonable if one takes a look in Figure 39 at how often Audi Neckarsulm’s demand for a certain week stated in the final supply call-off exceeded the demand for the week in which it announced the supply call-off four weeks earlier. If companies in this supply chain were ordering too much on purpose in one or several periods, they would have an excessive amount of inventory of parts and/or products for at least some time. As shown in Section 4.2.2, Wahler and TCG Systemtechnik have a relatively high average inventory level of thermostats for a certain amount of time in the monitored period. TCG Systemtechnik’s inventory level of water pumps is rather low on average, and high at times. Audi Hungaria’s inventory level of water pumps and engines is unknown. Furthermore, if excessive amounts were ordered in some periods, there must be periods without an order so that inventory is eventually reduced again. As shown in Table 18, Audi Hungaria and TCG Systemtechnik do not place an order in four periods. However, based on the available information, it cannot be determined what the reason for this is.

128

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

Alone, neither of these four explanations is likely to be sufficient to explain the difference between the number of engines ordered by Audi Neckarsulm, and the number of water pumps ordered by Audi Hungaria as well as the amount of thermostats ordered by TCG Systemtechnik. Probably, a combination of these four items (and perhaps others) causes the differences in the demanded quantities. However, due to a lack of information, no definite conclusion can be made for the reason why there is an increase in total demand and mean demand upstream in the engine supply chain. Irrespective of why total and, thus, also mean demand increase from echelon to echelon in this supply chain, the question that has to be answered is where all the water pumps and thermostats are if they have not been built into an engine delivered to Audi Neckarsulm. As can be seen in Table 18, Audi Neckarsulm received 1,586 engines, but Audi Hungaria got 2,721 water pumps to build these engines. Furthermore, TCG Systemtechnik received 3,934 thermostats from Wahler to make the mentioned amount of water pumps (plus 51 for VW’s location in Kassel). One speculation on the whereabouts of the thermostats, water pumps, and engines could be that the surplus is stored in a warehouse. On June 29, 2003 (i.e. by the end of week 26), TCG Systemtechnik had indeed a high inventory level of thermostats: 788 units (Gisela Kastner, personal communication, July 07, 2003). TCG Systemtechnik’s inventory level of water pumps on that day was zero. By that date, it had delivered 3,072 water pumps to other companies (Gisela Kastner, personal communication, August 04, 2003b); and from this, it follows that there must have been a work-in-transit inventory of 74 thermostats96 on June 29, 2003 (Gisela Kastner, personal communication, August 08, 2003). According to Kastner (personal communication, July 07, 2003), TCG Systemtechnik delivered 2,721 water pumps to Audi Hungaria and 51 to VW’s location in Kassel; thus, 2,772 units to the only two customers for that product (Gisela Kastner, personal communication, August 04, 2003a). The difference between 3,072 water pumps that were delivered by June 29, 2003 according to Kastner (personal communication, August 04, 2003b), and 2,772 water pumps that were delivered by June 29, 2003 according to Kastner (personal communication, July 07, 2003) might be explained as follows: Some water pumps were sent back by Audi Hungaria and had to be replaced. Thus, the cumulative amount of water pumps delivered to Audi Hungaria is actually larger than 2,721 which might account for the difference (Gisela Kastner, personal communication, August 08, 2003). If this speculation is correct, it would mean that 300 engine, i.e. ten percent, were sent back in the monitored period

96

Perhaps, work-in-transit inventory was even larger because it can be assumed that it was not necessary to build a new thermostat into every water pump that has been sent back. The work-in-transit inventory has been calculated as follows: 3,934 – 788 – 3072 = 74.

4.2 Description and Analysis of the Supply Chain of the Audi A8 V8 4.0l Diesel Engine

129

of time which can be considered much. In any case, this would be more than the “very little” scrap rate stated by Kastner (personal communication, August 08, 2003). Audi Hungaria delivered 1,586 engines (and, thus, water pumps) to Audi Neckarsulm, but it received 2,721 water pumps from TCG Systemtechnik. It is unclear where the other 1,135 water pumps are that had been sent to Audi Hungaria by TCG Systemtechnik. Possibly, the water pumps are sitting in Audi Hungaria’s storage area for water pumps, or they are work-in-transit inventory, or they were already built into engines which are now stored in Audi Hungaria’s engine storage area; although according to Szorady, no engines are stored in Györ (personal communication, July 16, 2003). These assumptions cannot be verified because the necessary information is not available. A second speculation on the whereabouts of the water pumps, and of the engines, could be that a substantial amount of the parts were of substandard quality and needed to be scrapped by their manufacturer. Kastner (personal communication, August 08, 2003) said that at TCG Systemtechnik the scrap rate is low. For the other companies, no information on the scrap rate is available; however, it seems unlikely that the scrap rate at these companies is high enough to explain the difference between orders of engines, of water pumps, and of thermostats. A third speculation on the whereabouts of the thermostats, water pumps, and engines could be that at least part of the surplus simply disappeared (e.g. because they have been stolen). To some extent, this seems to be corroborated by informal conversations held by the author with staff members at Audi; however, it could not be confirmed. Based on the available information, neither of the three discussed items can be proven to be either true or false and, therefore, they have to remain speculation. This ends the description and the analysis of the engine supply chain, and now it will be determined whether or not the bullwhip effect can be diagnosed in this supply chain.

4.3 Evaluation of the Audi AG Supply Chain 4.3.1 The Bullwhip Effect in the Supply Chain of the Audi A8 V8 4.0l Diesel Engine In this section, the engine supply chain will be examined as to the extent of the bullwhip effect. The basis for all calculations is demand data of this supply chain’s members between week 10 and week 26 of 2003 which was collected between July 04, 2003 and August 13, 2003 and was described in Section 4.2.3.

130

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

Table 19 shows that not only the total demand and, thus, also the mean demand increased from echelon to echelon, but also the variation of demand. This might indicate that the bullwhip effect occurs in this supply chain. The bullwhip effect in this supply chain with four levels97 was calculated by using the data from Table 18. Table 19: Quantification of the Bullwhip Effect in the Engine Supply Chain between Week 10 and Week 26 of 2003 Audi Neckarsulm

Audi Hungaria

TCG Systemtechnik

-

4.764

6.354

Bullwhip Effect

(Source: Based on data extracted from BETA93 (list LUA4000A) on August 01, 2003; Gisela Kastner, personal communication, July 07, 2003;Katalin Szorady, personal communication, August 13, 2003)

The data in Table 19 suggests that variation of demand between week 10 and week 26 of 2003 at Audi Hungaria is 4.764 times larger (i.e. 376.4 percent) than at Audi Neckarsulm. At TCG Systemtechnik, it is 6.354 times larger (i.e. 535.4 percent) than at Audi Neckarsulm. In other words, there is an immense increase in demand order variability. Therefore, it can be stated that the bullwhip effect is present in this supply chain to a large degree. Figure 40 visualizes demand order variability in the engine supply chain. 500 450 400 350

quantity

300 250 200 150 100 50 0 week 10

week 11

week 12

week 13

week 14

week 15

week 16

week 17

week 18

week 19

week 20

week 21

week 22

week 23

week 24

week 25

week 26

time Audi Neckarsulm

Audi Hungaria

TCG Systemtechnik

Figure 40: Demand Order Variability in the Supply Chain for the V8 4.0l Diesel Engine (Source: Based on data extracted from BETA93 (list LUA4000A) on August 01, 2003; Gisela Kastner, personal communication, July 07, 2003; Katalin Szorady, personal communication, August 13, 2003)

97

In Table 19, only the OEM, tier 1, and tier 2 are included. Tier 3 (Wahler) is not considered because Wahler’s orders to its suppliers were not included in the examination study.

4.3 Evaluation of the Audi AG Supply Chain

131

This figure shows strikingly why variation in demand is much larger at Audi Hungaria and at TCG Systemtechnik than at Audi Neckarsulm which results in the bullwhip effect: The graphs of Audi Hungaria and TCG Systemtechnik oscillate much more than Audi Neckarsulm’s. As shown in Section 2.2.4.2, such demand order variability, i.e. the bullwhip effect, can be caused by several factors. In the following paragraphs, the factors that might contribute to the bullwhip effect in the engine supply chain will be discussed. Demand Forecast Updating According to Lee et al. (1997b), one cause is “demand forecast updating”; however, this seems to have little significance in this particular supply chain because no member of this chain seems to produce based on forecasts. Instead, each company only manufactures (or at least tries to manufacture) what their customer ordered by supply call-off (Gisela Kastner, personal communication, July 04, 2003a; Jörg Binick, personal communication, July 07, 2003; Katalin Szorady, personal communication, July 16, 2003; Markus Bedic, personal communication, July 04, 2003). Order Batching A second cause identified by Lee et al. (1997b) is “order batching”. As described in Section 2.2.4.2, four factors contribute to this practice: the pursued inventory policy, transport optimization, lot size optimization, and sales quotas/incentives. Transport optimization (e.g. postponing transportation until a full truckload of goods has been accumulated) does not seem to play a role because all companies in the engine supply chain outsource transportation. Sales quotas or similar incentives are also not relevant in the engine supply chain. And as will be shown in the following paragraphs, neither of the two other factors seems to apply to the situation in the engine supply chain. Therefore, order batching as described in Section 2.2.4.2 does not seem to be the cause of the bullwhip effect in the engine supply chain. According to its inventory policy, Audi Neckarsulm places orders of varying amounts in each period (see Table 18 and Figure 40). Its small lot size (six engines)98 enables this company to order relatively precisely what it needs in a particular period and, therefore, it is never in a situation where it has enough engines in stock to manufacture the A8s with a V8 4.0l diesel

98

There are two reasons why the amounts shown as orders in Table 18 are not multiples of six, the lot size for the engines. First, what is shown in this table are not orders, but deliveries of engines as an approximation to actual orders. Possibly, the amount received and the amount ordered deviate and the latter amount was a multiple of six. Second, there are backlogs and immediate demand in addition to “regular” demand. These two positions are not optimized. Only “regular” demand is lot size optimized. As a result, total demand might not be a number that is a multiple of six.

132

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

engine demanded in a particular period without ordering engines. Therefore, it can be assumed that variation of demand at Audi Neckarsulm is not caused by order batching because its inventory policy allows it to order in each period and its lot size is quite small. Variation in Audi Neckarsulm’s demand is probably caused by varying end customer demand. As the following paragraph will show, Audi Hungaria’s variation in demand is also not caused by order batching. According to its inventory policy, the (t,S)-policy, it can place an order in each period (i.e. each week) to replace those water pumps which were used during the previous week for manufacturing engines.99 Therefore, the implemented inventory policy does not contribute to order batching at this level of the engine supply chain. In four periods, Audi Hungaria does not order anything which might be explained with the pursuit of lot size optimization: If companies aim at always ordering in lot sizes, they order more than they actually need in some periods and, hence, they do not need to order anything in subsequent periods. As a result, variation in demand is great. However, Audi Hungaria’s order pattern can hardly be explained by this line of reasoning and, thus, not by order batching due to lot size optimization. As Table 18 shows, this company usually orders more than it needs in a particular period, but a period with an order that exceeds demand is rarely followed by one or more periods with no or only small demand. The assumption does not seem too far-fetched that this company must have a large inventory of water pumps and/or of engines by the end of week 26 of 2003: If the inventory level of water pumps was zero at the beginning of week 10 of 2003, there must be a total of 1,135 water pumps100 which are either stored in the storage area for parts, or which have already been built into engines and which are now stored in the respective storage area, or which are work-in-transit inventory. This assumption cannot be verified because no inventory data was provided by Audi Hungaria. The question is why does Audi Hungaria usually place orders that seem too large. Based on the available information, there is no rational reason for this, and the responsible people were not available for providing an explanation. Variance of Audi Hungaria’s demand is large because this company orders either zero water pumps or varying multiples of its lot size 150 (or quantities that are close to that amount). However, because of what has been described in the above paragraph, the reason for the great variation in this company’s orders is not order batching as described in Section 2.2.4.2 despite the

99

According to this policy, no water pumps need to be ordered if no engines are produced. However, this is not the case in the engine supply chain: Four times, Audi Hungaria does not order water pumps; although it used some to manufacture engines.

100

This number represents the difference between received water pumps and delivered engines.

4.3 Evaluation of the Audi AG Supply Chain

133

fact that Audi Hungaria orders in batches of about 150. Furthermore, Figure 40 clearly shows that demand quantities by Audi Hungaria vary much more than demand quantities by Audi Neckarsulm. Hence, variation in demand at Audi Hungaria is much larger than variation in demand at Audi Neckarsulm. Based on the available information, it cannot be explained logically why Audi Hungaria orders vary to a large degree in almost every period. However, this practice is the reason why Audi Hungaria’s variation in demand is so large and why it exceeds Audi Neckarsulm’s variation in demand. The inventory policy pursued by TCG Systemtechnik, the (t,S)-inventory policy, is also not sufficient to explain variation in this company’s demand because this policy allows the company to place an order every period. Moreover, order batching due to lot size optimization is hardly a suitable argument for explaining the large variation in demand at TCG Systemtechnik: TCG Systemtechnik did not have a lot size for ordering thermostats until June 2003, and it ordered just what it needed; since then, its lot size is 192 pieces (Gisela Kastner, personal communication, August 08, 2003). Therefore, its orders should closely reflect demand in a certain period. Figure 40 visualizes the reason why variation in demand is larger at TCG Systemtechnik than at Audi Hungaria: Most of the time, TCG Systemtechnik ordered more than Audi Hungaria, thus, the number of periods with no order is the same at these two levels of the engine supply chain. The question is why did TCG Systemtechnik order more thermostats than it needed to produce the demanded water pumps. One reason why TCG Systemtechnik’s variation in demand is larger than variation in demand at any other level of the engine supply chain might be the following: Audi Neckarsulm, and probably also Audi Hungaria, announced high demand for a certain week in the future; but less than four weeks before delivery was due, they lowered demand for that week again while TCG Systemtechnik did not have this option since the first four weeks of its supply call-off to Wahler are fixed. As a result, TCG Systemtechnik was forced to order a large amount; even though it did not need this quantity anymore and probably reduced or even canceled future orders until the surplus was depleted. Hence, variation in demand is large. According to Kastner (personal communication, August 08, 2003), a sudden decrease in demand by Audi occurred in particular in June 2003. Therefore, one might conclude that sudden decreases in Audi Hungaria’s demand combined with the lack of reaction scope by TCG Systemtechnik are the reason for the fact that between March and June 2003 variation in demand at TCG Systemtechnik is larger than variation in demand at Audi Hungaria. However, as Figure 39 shows Audi Neckarsulm (and, thus, probably also Audi Hungaria) increased demand just as often as it decreased demand. Therefore, the increase of TCG

134

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

Systemtechnik’s variation in demand when compared to Audi Hungaria’s might also be explained as follows: Since TCG Systemtechnik feared that Audi Hungaria would suddenly increase its order quantity for a certain period, it ordered surplus thermostats as a safety buffer. If Audi Hungaria did not increase its demand for a particular week in the final supply call-off or in one of the last supply call-offs, TCG Systemtechnik had surplus thermostats (or water pumps) and, thus, could lower its orders to Wahler for the upcoming periods. This practice would also result in large variation in demand. A third reason why TCG Systemtechnik seems to order too much might simply be irrational order behavior: This company might order too much for no apparent reason just like Audi Hungaria seems to do. However, it has to be emphasized that this statement is only an assumption which cannot be verified in this dissertation.101 Based on the available information, it cannot be determined which of the suggested reasons causes the great variation in TCG Systemtechnik’s demand. All that can be stated is that it is probably not caused by order batching as described in Section 2.2.4.2. Price Fluctuations The third cause of the bullwhip effect identified by Lee et al. (1997b), “price fluctuations”, does not apply in this supply chain because the prices for engines, water pumps, and thermostats are constant over a longer period of time. Prices are constant since they have been fixed in the contracts between Audi Neckarsulm and Audi Hungaria, between Audi Hungaria and TCG Systemtechnik as well as between TCG Systemtechnik and Wahler. Rationing and Shortage Gaming The fourth possible cause of the bullwhip effect mentioned by Lee et al. (1997b) is “rationing and shortage gaming”. However, neither rationing nor shortage gaming are likely to be an issue in this supply chain because all relationships in the engine supply chain are one-to-one customersupplier and supplier-customer relationships (Gisela Kastner, personal communication, July 04, 2003a; Markus Bedic, July 04, 2003; Jörg Binick, personal communication, July 07, 2003; Katalin Szorady, personal communication, July 16, 2003).102

101

Based on the impressions the author gained of TCG Systemtechnik, this kind of behavior seems rather unlikely.

102

The only exception is that TCG Systemtechnik also delivers a few water pumps to the VW Group’s central spare parts warehouse in Kassel. However, this should not have an impact on the amount Audi Hungaria gets because, as has been stated, VW in Kassel only receives a very small amount compared to the quantities that are delivered to Audi Hungaria.

4.3 Evaluation of the Audi AG Supply Chain

135

This section has shown that the bullwhip effect is present in the engine supply chain. It will now be evaluated if the cooperation scenarios identified in Chapter 3 could be implemented and be beneficial to the supply chain of the Audi A8 V8 4.0l diesel engine so as to alleviate the bullwhip effect and its detrimental effects on the entire supply chain and its members.

4.3.2 Evaluation of the Cooperation Scenarios for Transportation in the Supply Chain of the Audi A8 V8 4.0l Diesel Engine The evaluation of the scenarios is based on the assumption that orders are only made if the goods can be transported efficiently. As has been pointed out in Section 3.1.1, transporting freight in truckloads is usually the least expensive. However, as has also been pointed out in Section 3.1.1, ordering truckloads of goods comes with the disadvantage of high inventory levels. In other words, there is a trade-off between inventory carrying costs and transportation costs. The solution in which transport costs equal inventory carrying costs can be regarded as optimal. The order size that leads to this optimal solution of the trade-off is considered in this dissertation as the efficient order size and the respective shipping volume leads to efficient utilization of the chosen means of transportation. In Section 3.1.2, scenarios were introduced that describe how members of a supply chain can reap various benefits by cooperating in the field of transportation. It will now be contemplated if these scenarios could be implemented and be beneficial in the analyzed Audi supply chain if they have not already been put into practice.

4.3.2.1 Cooperation Scenario I: Engaging in a Logistics Alliance Lee et al. (1997b) claim that order batching is one of the major causes of the bullwhip effect and as was shown in Section 2.2.4.2, one of the reasons for this practice is the effort to ensure the efficient transportation of goods: Regarding transportation costs, it might only be efficient to order large batches (e.g. full truckloads). This might result in great variation in demand because a large order in one period is likely to be followed by one or several periods without an order – depending on the demand for the company’s end product. If order batches upstream in the supply chain are larger than downstream, variation in demand upstream is probably larger than downstream, and this results in an extensive bullwhip effect. In such a constellation, Cooperation Scenario I would be beneficial, i.e. it would facilitate an alleviation of the bullwhip effect if it contributed to reducing the size of an efficient order batch so that a small order can be placed efficiently in every period and if the decrease of variation in demand is larger upstream in the supply chain than downstream. This cooperation scenario might actually help reduce the size of an efficient order batch because an LSP can pool freight from different customers and, thus,

136

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

achieve an efficient level of utilization of the truck, airplane, etc.; although each of its customers only has a small amount of freight to be transported. In the constellation described above, in which Cooperation Scenario I has not been implemented, a customer only places an order if this order has a certain size so that it can transport the freight efficiently. It is also possible that the supplier bundles several orders from its customer until enough has been accumulated to efficiently transport the goods. In this second constellation, implementing Cooperation Scenario I might result in a shorter lead time because this scenario might allow an efficient increase in the frequency of transports between two companies in a supply chain due to the pooling of freight. In Section 2.2.4.2, it was stated that reducing lead time might result in a lower bullwhip effect (see also (5.23) in Section 5.1.3.1). Therefore, one might state that this scenario would be beneficial, i.e. it would contribute to alleviating the bullwhip effect if the second constellation applies, should it facilitate a shortening of lead time, and if variation in demand is decreased to a larger extent upstream than downstream. Since this constellation does not apply in the engine supply chain, it will not be discussed any further. As described in Section 4.2.1, Audi Neckarsulm and Audi Hungaria use VW T as a long-term LSP, or more precisely, as a transport broker (see Section 3.1.1). VW T purchases train transportation services from DB Cargo AG and truck transportation services from Schachinger for the transportation of water pumps to Audi Hungaria, and for the transportation of engines to Audi Neckarsulm. Therefore, one can state that DB Cargo AG and Schachinger both serve as long-term LSPs for VW T. Schenker AG acts as an LSP for TCG Systemtechnik (see Section 4.2.1): On behalf of TCG Systemtechnik, it transports the thermostats from Wahler to TCG Systemtechnik. Thus, it can be stated that there is a logistics alliance as defined in Section 3.1.2.1 between DB Cargo AG and VW T, between Schachinger and VW T, between VW T and Audi Neckarsulm, between VW T and Audi Hungaria, and, last but not least, between TCG Systemtechnik and Schenker AG. In other words, this cooperation scenario has already been implemented by the members of the engine supply chain. The fact that this scenario is already implemented in the engine supply chain probably has the effect that each company’s variation in demand is smaller than it would be without the logistics alliance. The reason for this is the following: If the members of this supply chain were not in a logistics alliance for transportation with an LSP, i.e. if they were insourcing transportation, their efficient order size would probably be larger than the efficient order size when outsourcing transportation because when insourcing, they cannot consolidate their freight with the freight of other companies in order to achieve a large volume to be transported. As a result, there would probably be a rather large order in one period, followed by several periods without an order and,

4.3 Evaluation of the Audi AG Supply Chain

137

thus, variation in demand would be large. For example, Audi Hungaria’s average weekly demand quantity in the monitored period of time was about 160 water pumps. This amount of freight is probably not enough for efficient transportation if it was transported on its own. However, since the LSP pools the water pumps with other freight, it can efficiently transport this rather small amount of water pumps. If Audi Hungaria were transporting the water pumps on its own, it would probably need to postpone an order until the demand quantity is large enough to ensure efficient transportation (or it would have to accept excessive unit costs for transportation which can be regarded as inefficient). That means that Audi Hungaria would probably not be able to place a rather small order each week as it is currently doing; but it would have to make a large order in one week, and it would not be able to order anything for several weeks after that. As a result, its variation in demand would be immense. The same line of thought can be applied to the other members of this supply chain. Therefore, it can be stated that the implementation of Cooperation Scenario I might have contributed to keeping the variation in demand of each company in the engine supply chain at a lower level than it would have been otherwise. The bullwhip effect in the engine supply chain is possibly kept at a lower level than it would have been otherwise if this scenario has helped to reduce the variation in Audi Hungaria’s demand and in TCG Systemtechnik’s demand to a greater extent than the variation in Audi Neckarsulm’s demand. The picture might look different, however, if lead time is also factored into the equation: By implementing this scenario, transportation time and, thus, total lead time might possibly be increased which might result in a larger bullwhip effect (see Section 2.2.4.2). The reason for this is that an LSP usually does not deliver the freight directly. For example, Schenker AG does not transport the thermostats directly from Wahler to TCG Systemtechnik, but first brings them to one of its depots in Southern Germany where the thermostats are possibly stored for some time and then loaded on a different truck together with other freight. This truck is bound for Schenker AG’s depot in Linz (Austria) where the thermostats are reloaded on yet another truck before finally being delivered to TCG Systemtechnik (Markus Bedic, personal communication, July 04, 2003). If the truck went from Wahler straight to TCG Systemtechnik, transportation lead time could possibly be reduced from three days to one day because the distance between Esslingen and Micheldorf is not too large (576 kilometers). Another instance of transportation lead time being increased due to using an LSP and due to trying to consolidate freight is the transportation of water pumps from TCG Systemtechnik to Audi Hungaria. As already described, the water pumps are first brought to Ingolstadt by truck, and then loaded on a train to Györ. This takes four days; however, if the water pumps were delivered directly, transportation time could be cut

138

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

to one day, or about six hours to be more precise (Katalin Szorady, personal communication, July 16, 2003). As a conclusion, it can be stated that this cooperation scenario might help lower the bullwhip effect and, thus, it might be advantageous to the supply chain and its members. However, this is only true under two conditions: 1. Efficient order sizes upstream in the supply chain must be reduced to a greater extent than downstream so that variation in demand upstream is reduced more than variation in demand downstream. 2. Lead time might possibly be extended if this scenario is implemented which might increase the bullwhip effect. This increase of the bullwhip effect due to longer lead time is not to cancel out the reduction of the bullwhip effect due to smaller, efficient order batches. This applies to each supply chain, in general, as well as to the engine supply chain, in particular.

4.3.2.2 Cooperation Scenario II: Supply Chain-wide Container Management The implementation of this scenario might be beneficial in terms of alleviating the bullwhip effect and its detrimental effect on the supply chain as a whole as well as on its individual members if it helps to eliminate the necessity to repack goods while they are shipped between two points in the supply chain. Repacking takes time and, thus, increases transportation lead time which might contribute to an increase of the bullwhip effect. The transportation of the thermostats from Wahler to TCG Systemtechnik is done by truck. No repacking is necessary because there is no switching between different means of transportation and all transportation is in the hands of one party, namely Schenker AG, functioning as LSP (Gisela Kastner, personal communication, July 04, 2003a; Markus Bedic, personal communication, July 04, 2003). Therefore, this scenario does not bear any potential for reducing total lead time and, thus, the bullwhip effect between these companies. For transporting the water pumps from TCG Systemtechnik to Audi Hungaria, two modes are used, truck and train, but the freight is only reloaded, not repacked when modes are switched (Katalin Szorady, personal communication, July 16, 2003). Consequently, it can be said that some sort of container management is practiced. This practice might have helped to keep lead time and, thus, also the bullwhip effect at a lower level than it would have been otherwise. Transportation between Audi Hungaria and Audi Neckarsulm is done by train; unless it is a case of expedited shipping. Similarly to the transportation between Wahler and TCG Systemtechnik as well as between TCG Systemtechnik and Audi Hungaria, no repacking takes place (Jörg Binick,

4.3 Evaluation of the Audi AG Supply Chain

139

personal communication, July 07, 2003). Therefore, this scenario does not bear any potential for reducing lead time and, thus, the bullwhip effect between these two companies. As a conclusion, it can be stated that this scenario might be beneficial for the supply chain and its members because it might bear some potential for reducing the bullwhip effect and, thus, its negative impact on the supply chain as a whole as well as on its members since it might help to shorten lead time. However, this is only the case if variation in demand upstream in the supply chain is reduced to a greater extent than downstream by implementing this scenario. In the sample supply chain, this scenario might have helped to keep lead time between TCG Systemtechnik and Audi Hungaria lower than it would have been otherwise; and, thus, it might have contributed to keeping variation in Audi Hungaria’s demand at a lower level than it would have been otherwise. Consequently, the bullwhip effect between Audi Neckarsulm and Audi Hungaria might also be lower than it would have been otherwise and, therefore, this scenario can be regarded as beneficial for this part of the supply chain. For the other parts of this supply chain, this scenario is not of any relevance and, thus, does not bear any improvement potential.

4.3.2.3 Cooperation Scenario III: Selling Excess Transportation Capacity to Other Companies It goes without saying that a company can only sell excess transportation capacity to another if it does not outsource this function. A further prerequisite is that the seller and the buyer of the transportation capacity are located in the same area, or that the buyer is located on or near the route between the supplier offering to sell its excess capacity to its customer. Supposing that these requirements are met and under the condition that variation in demand is decreased upstream in the supply chain to a greater extent than downstream, the bullwhip effect and its negative consequences might be alleviated by implementing this scenario. This is the case if this scenario allows a company to maintain efficiency of transportation and to still order a small batch in each period instead of ordering a large quantity in one period and nothing in subsequent periods which is one of the major causes of the bullwhip effect (see Section 2.2.4.2). A company might be able to efficiently order (and also transport) small amounts in each period if this scenario is implemented because the company’s small amount of freight is pooled with freight of other companies. Thus, an efficient utilization of the truck or other means of transportation can be guaranteed; although the freight of each individual company is only of a small quantity. Since all members of the analyzed engine supply chain outsource their transportation function, it is not possible to implement this scenario in the sample supply chain and, thus, this scenario does not bear any potential for reducing the bullwhip effect in this particular supply chain.

140

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

4.3.2.4 Cooperation Scenario IV: Joint Ownership of Transportation Capacity Here, the same reasoning and requirements as in Scenario III apply. Additionally, such form of cooperation only seems feasible between two or more companies that plan to work together for a longer period of time (e.g. the lifetime of the transportation means) in order to avoid the issue of splitting up the property. Although this scenario might generally be beneficial to a supply chain and its members regarding the bullwhip effect, it does not bear any potential for the engine supply chain because all of its members outsource their transportation function. However, if the VW Group companies were not outsourcing their transportation needs, such scenario would be conceivable between any company of this group, for example between Audi Neckarsulm/Audi Ingolstadt and Audi Hungaria, because a long-term relationship between these companies is very likely as they belong to the same group. The companies involved in this scenario could possibly profit from a lower bullwhip effect.

4.3.2.5 Cooperation Scenario V: Multi-Stop Shipping and Sequenced Loading A requirement for this scenario is that multiple suppliers deliver to one manufacturer. Another requirement is that the involved suppliers are located at relatively close proximity to each other or on the same route to the manufacturer. Supposing that these requirements are met and under the condition that variation in demand upstream in the supply chain is reduced to a greater extent than downstream, the bullwhip effect and its negative consequences might be alleviated through implementing this scenario because this scenario might enable a company to keep up efficient transportation and allow it to still order a small batch in each period instead of ordering a large quantity in one period and nothing in subsequent periods which is one of the major causes of the bullwhip effect (see Section 2.2.4.2). Efficiently ordering (and also transporting) small amounts in each period might be achieved by a company through implementing this scenario because for transportation its freight is pooled with the freight of other companies. This way an efficient utilization of the truck or other means of transportation can be guaranteed; even though the freight which each individual company wants to transport is possibly only little. A further reduction of the bullwhip effect might be attained by reducing lead time through sequenced loading. This technique allows a shortening of the time between receiving a part and being able to use it: The receiver of the parts does not need to put them in the correct sequence for assembly anymore. It has already been mentioned that reducing lead time might result in a lower bullwhip effect which is also expressed in Formulas (5.23) and (5.24).

4.3 Evaluation of the Audi AG Supply Chain

141

Neither of the above mentioned requirements are met in the engine supply chain and, therefore, it does not seem possible to implement this cooperation scenario. As a consequence, this scenario does not bear any improvement potential for this supply chain.

4.3.2.6 Cooperation Scenario VI: Merge-in-Transit and Sequenced Loading A requirement for this scenario is that multiple suppliers deliver to one manufacturer. If this requirement is met and if variation in demand upstream in the supply chain decreases more than downstream, this scenario might be beneficial for a supply chain and its members because it might facilitate a reduction of the bullwhip effect and its negative consequences. This might be the case because this scenario might allow a company to maintain transportation efficiency and still allow it to order a small batch in each period instead of ordering a large quantity in one period and nothing in subsequent periods which is one of the major causes of the bullwhip effect (see Section 2.2.4.2). A company might be able to efficiently order (and also transport) small amounts in each period if this scenario is implemented because its freight is pooled at a merge center with freight of other companies for transportation. This way, an efficient utilization of the truck or other means of transportation can be guaranteed for at least the largest part of the distance that is to be covered between the suppliers and their customer; although the amount of freight that each individual supplier ships might only be small. By sequenced loading, a further bullwhip effect reduction might be attained by shortening lead time (see Section 2.2.4.2) because sequenced loading allows one to shorten the time between receiving a part and being able to use it: It is no longer necessary to put the parts in the right sequence. In the engine supply chain, the above mentioned requirement for implementing this scenario is not met and, therefore, it does not seem possible to implement this cooperation scenario. As a consequence, this scenario does not bear any improvement potential for the analyzed supply chain.

4.3.2.7 Cooperation Scenario VII: Cross Docking and Sequenced Loading This scenario requires multiple suppliers to deliver the same set of parts (or at least a similar one) to the same manufacturers or to the same plants of one manufacturer. If this scenario is implemented, each supplier brings a full truckload of its parts to the cross docking center where its parts are mixed with the parts of other suppliers. Several fully loaded trucks leave the cross docking center containing a mix of parts from all suppliers bound for different manufactures that all need the same or a similar mix of parts (see Section 3.1.2.7). In other words, this scenario does not aim at pooling freight which would allow each participant to efficiently order and transport

142

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

small amounts; and, therefore, it does not tackle the problem “order batching” which is one of the major causes of the bullwhip effect (see Section 2.2.4.2). This scenario also does not tackle any other issue that causes the bullwhip effect. Therefore, it can be stated that this scenario probably does not bear any potential for reducing the bullwhip effect and its negative impact on the supply chain as a whole as well as on its individual members. This scenario cannot be put into practice in the analyzed critical path of the engine supply chain because the above mentioned requirement for its implementation is not met.

4.3.2.8 Cooperation scenario VIII: Sequenced delivery By implementing this scenario, the time span between ordering and being able to assemble a part might be reduced since the intermediate steps of storing the parts, checking their quality, and putting them into the right order are omitted (Thomas Müller, personal communication, July 02, 2003). Therefore, lead time might be shortened and, thus, variation in demand might decrease. If variation in demand is decreased upstream in the supply chain to a greater extent than downstream, the bullwhip effect is remedied and, consequently, its adverse consequences for the supply chain as a whole and for its individual members as well (see Section 2.2.4.2). Hence, this scenario might be beneficial. As already described in Section 4.1.1.2, this form of delivery is often used for parts characterized by bulkiness, complexity, and a high number of variants. However, neither the thermostat nor the water pump meet these requirements and, therefore, it seems logical that they are not delivered just in sequence (Gisela Kastner, personal communication, July 04, 2003a; Katalin Szorady, personal communication, July 16, 2003; Markus Bedic, personal communication, July 04, 2003). Therefore, this scenario does not bear any potential for reducing lead time in this part of the supply chain and, thus, also for reducing the bullwhip effect as well as its negative consequences. The engines delivered from Audi Hungaria to Audi Neckarsulm could be considered as complex, bulky parts with a number of variants.103 Therefore, delivering the engines for the A8, including the V8 4.0l diesel engine, just in sequence could be an option (Jörg Binick, personal communication, July 07, 2003). The question is if this can be put into practice because different types of engines require different frames for transportation due to their different sizes. Another obstacle to sequenced delivery of the engines from Györ to Neckarsulm might be the large distance that needs to be covered between the two plants. It can be considered a bold venture to

103

For example, engines which are built into cars to be delivered to cold countries have an isolation layer while others might not have this extra layer if they are built into cars to be delivered to warm countries. Furthermore, emission values might also vary depending on the car’s country of destination.

4.3 Evaluation of the Audi AG Supply Chain

143

try to deliver the engines just in sequence over that distance because much can happen while the engines are in transit for two days, e.g. the production plan might be changed. These two issues make it unlikely that this scenario can be implemented successfully between Audi Neckarsulm and Audi Hungaria. As a conclusion, it might be stated that this scenario might be beneficial for a supply chain and its members because it might potentially reduce the bullwhip effect. However, a lower bullwhip effect will only result from implementing this scenario if variation in demand upstream in the supply chain is decreased to a greater extent than downstream. For the engine supply chain, this scenario does not bear any improvement potential and can, consequently, not be regarded as beneficial for this particular chain.

4.3.3 Evaluation of the Implementation of Supply Chain Monitoring in the Audi AG Supply Chain: Real-time Exchange of Information on Capacity, Inventory, and Demand As mentioned in Section 3.2.1, supply-chain-wide monitoring is currently being tested in the engine supply chain within the GEKO project using ICON-SCC as a supporting tool and, therefore, the purpose of this section is to discuss potential benefits of its realization. By utilizing ICON-SCC, all participating companies have the opportunity to see available production capacity and its current utilization, inventory levels as well as gross demand of other supply chain members in the current period. In other words, probably more relevant information than before is exchanged between companies in real-time. However, the answers given by the interviewed people showed that even before using supporting tools, “neighboring” companies in the chain had some knowledge of the other companies. For example, before piloting ICON-SCC, TCG Systemtechnik had information on Wahler’s capacity and lead time. From Audi Hungaria, it knew the approximate net demand for an entire year and additionally, it had the information from the weekly supply call-offs providing an outlook on Audi Hungaria’s demand for the upcoming six months (Gisela Kastner, personal communication, July 04, 2003a). Parallel to this, Wahler knew roughly TCG Systemtechnik’s net demand for an entire year, and it got supply calloffs on a weekly basis which provided information on TCG Systemtechnik’s net demand for the upcoming six months whereby the demand data for the first four weeks was legally binding (Markus Bedic, personal communication, July 04, 2003). Audi Hungaria did not get an outlook on demand for the upcoming year, but on a weekly basis it received supply call-offs with information on Audi Neckarsulm’s anticipated demand for the next six months (Jörg Binick, personal communication, July 07, 2003; Katalin Szorady, personal communication, July 16, 2003).

144

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

Without ICON-SCC, most companies were presumably missing information on the other companies’ capacity and its utilization, inventory levels, and gross demand. Therefore, one can conclude that the value added by using ICON-SCC and similar tools is that this information becomes available to members of a supply chain like the analyzed one. It now needs to be contemplated what the advantage of this scenario is, i.e. to which extent extra information can help to mitigate the bullwhip effect in a supply chain. When this scenario is implemented, a company in a supply chain has access to information on capacity and its utilization, inventory levels, and gross demand of other supply chain members. Using a software tool is not the only way to distribute this information in a supply chain, but a customer can also contact its supplier by phone, fax, email, etc., or simply ask. It might not get information like, “Next week my capacity utilization will be 75 percent.”; however, it probably does not even care about this kind of information. What a customer probably wants to know is if this supplier is able to deliver a certain amount at a certain date. And probably, a supplier would be willing to give this kind of information to the customer (Jörg Binick, personal communication, July 07, 2003) because the more the supplier can sell to a customer, the more the supplier can earn from this one customer. This means that even without supporting software tools, the customer can get information indirectly on its supplier’s capacity utilization. Therefore, by only looking at the dyadic relationship between two “neighboring” companies in a supply chain, the benefits of software tools facilitating information exchange are not extensive regarding the exchange of capacity information. However, the advantages become more obvious when broadening the scope of the picture. For example, if Audi suddenly wants to produce more Audi A8s with a V8 4.0l diesel engine, Audi Hungaria needs to be able to increase its production of these engines, TCG Systemtechnik needs to be able to manufacture more water pumps, and Wahler needs to be able to produce more thermostats. Without a supporting software tool, each company needs to check with its supplier if the supplier is able to deliver the increased amount. Depending on the total number of suppliers involved, this is probably cumbersome and time consuming. With a supporting software tool, Audi has the capacity utilization of all critical suppliers of its engine supply chain directly available. For inventory levels, a similar reasoning applies as for the level of capacity utilization. In a dyadic relationship, a customer might be able to get inventory information from its supplier and the other way around – if both parties are cooperative. The advantages of software tools facilitating the real-time exchange of information throughout the supply chain once again become more obvious when broadening the scope of the picture: For example, Audi wonders if there are enough parts in the chain for a two percent increase in the production of Audi A8s with V8 4.0l

4.3 Evaluation of the Audi AG Supply Chain

145

diesel engines within a short period of time. With a supporting software tool, Audi can easily and instantly extract this information. It has to be emphasized that ICON-SCC, the software tool which Audi is currently testing, does not provide information on future demand and on future capacity utilization. Therefore, it cannot be used to check if there are enough resources in the pipeline to handle a demand increase on short notice. However, there might be tools that offer this feature. It enables the members of a supply chain to take a look into the future and to play with various demand planning scenarios differing in quantities demanded. Furthermore, if such a feature is implemented, each member can see the impact that its actions have on companies further upstream. In addition, the latter have the opportunity to plan ahead, basing their decisions on the source of the demand and not only in the advance information from their direct customer. However, since the advance information provided by Audi Neckarsulm seems not to be reliable (see Figure 39), this scenario might not be as beneficial as it could be. At first sight, the supply chain-wide exchange of demand information seems to be the biggest benefit in terms of alleviating the bullwhip effect that this scenario has to offer because the lack of information on future demand is a major cause of this effect. As described in Sections 2.2.4.2 and 3.2.1, the unavailability of this data forces companies to make forecasts of their customers’ demand which always distorts information on actual demand and, thus, leads to the bullwhip effect. In the dyadic customer-supplier relationships of the engine supply chain, suppliers do not need to produce based on forecasts of their customer’s demand because demand information is provided. Once a year, TCG Systemtechnik and Wahler get an outlook on their customer’s demand in the upcoming year and all companies except Audi Neckarsulm receive weekly supply call-offs from their customer. Like the other companies in this supply chain, Audi Neckarsulm does not make forecasts of customer demand, but it determines its demand for V8 diesel engines by asking its dealers and importers how many A8s with V8 diesel engines they expect to sell in the upcoming months (Herbert Hammer, personal communication, July 10, 2003; Kerstin Wildebör, personal communication, July 09, 2003). Therefore, the bullwhip effect in the engine supply chain is not caused by a distortion of demand information due to demand forecasting as described by Lee et al. (1997b). One possible benefit of implementing this scenario and of using supporting software tools might be that a different kind of demand information is exchanged with the one currently used: the exchange of gross demand instead of net demand. At the moment, the members of the engine supply chain inform their supplier about their net demand via weekly supply call-offs. As

146

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

described in Section 3.2.1, net demand deviates from gross demand which is the actual demand for parts.104 A supply chain-wide exchange of gross demand is advantageous to a supply chain if it results in a lower bullwhip effect compared to the current practice of exchanging net demand. Three constellations of exchanging gross demand are possible; whereby all might differ in their potential to reduce the bullwhip effect: •

Bilateral or supply chain-wide exchange of gross demand with local optimization of order quantities (Constellation 1),

•

bilateral or supply chain-wide exchange of gross demand with no optimization (Constellation 2),

•

and supply chain-wide exchange of gross demand with supply-chain wide optimization of order quantities (Constellation 3).

Constellation 1: Bilateral or Supply Chain-wide Exchange of Gross Demand with Local Optimization of Order Quantities Each system usually tries to optimize the activities, e.g. production, that it carries out; whereby a system can consist of either a single company, or of multiple companies forming a supply chain. In Constellation 1, local optimization is pursued, i.e. optimization at the level of the individual company. This means that each company, for example, tries to optimize its production by manufacturing in optimal lot sizes. Furthermore, it can mean that this party in a dyadic relationship, which is responsible for paying freight costs, tries to optimize transportation. And finally, restrictions, like the availability of containers, have to be considered as well. In addition to the planned production quantity, these local optimization efforts and restrictions determine the quantity ordered by a company from its supplier, i.e. the net demand (Stefan Mayer, personal communication, May 21, 2003). It follows from this that if gross demand were exchanged bilaterally, or supply chain-wide, with a tool like ICON-SCC; and if companies pursued local optimization, a supplier would still have to meet its customer’s net demand. Even if the supplier knows the gross demand of its customer and of all other supply chain members because the customer’s net demand, which is the result of this party’s optimization routines and of considering this party’s restrictions like the availability of containers, is still binding. To the supplier, gross demand of its customer and possibly of other supply chain members is just a piece of extra information that is received in addition to net demand; net demand is still the quantity that the supplier has to deliver to its customer –

104

For instance, if Audi Hungaria wants to build five engines, its gross demand for water pumps is five as well.

4.3 Evaluation of the Audi AG Supply Chain

147

regardless of gross demand. This additional information “gross demand” in combination with information on inventory levels might help a supplier to anticipate its customer’s upcoming net demand if reliable information on future net demand is not available in advance, and if the supplier has gross demand information from a large number of periods so that it has the chance to recognize certain patterns. If it can gain this additional knowledge on its customer based on gross demand, it might be in a better position to properly adjust its inventory, its capacity, and its future demand for parts. Furthermore, it might be able to reduce its safety stock which is a protection against sudden future demand increases. If there is less uncertainty due to more information, safety stocks as a substitute for information can possibly be lowered as well (Oliver, 1999). Based on this reasoning, one can state that this constellation of exchanging gross demand bilaterally or supply chain-wide might bear some potential for reducing variation in demand and, thus, possibly also the bullwhip effect if variation in demand upstream in the supply chain is reduced to a greater extent than downstream. The reason is that gross demand (plus inventory levels) as an additional information for net demand might enable each member of a supply chain to anticipate its customer’s future (net) demand and, thus, to properly adjust production of products and demand for parts – on the condition that gross demand shows a certain repetitive pattern over time. In the engine supply chain, the exchange of gross demand information is already implemented: ICON-SCC facilitates the exchange of this information. However, the advantage of exchanging gross demand is questionable because, as the data from BESI2 (extracted on July 26, 2003; see Appendix J) shows, Audi Neckarsulm’s gross demand for a certain week changes frequently. Audi Hungaria’s gross demand is not available; however, it seems likely that Audi Hungaria also frequently adjusts its demand for a certain week if Audi Neckarsulm does this as well. If gross demand changes are mirrored in net demand changes, gross demand of these two companies probably does not follow a pattern; but instead, demand appears to be adjusted randomly and abruptly. At least one instance of such abrupt change of demand occurred in the monitored period of time. Since a different kind of engine was temporarily unavailable due to supply shortages, Audi Neckarsulm decided to increase its demand for V8 4.0l diesel engines; and, consequently, Audi Hungaria’s demand for the respective water pump increased as did TCG Systemtechnik’s demand for the respective thermostat. Suddenly, however, after having announced a large demand for certain weeks, Audi Neckarsulm lowered its demand for the engine (Gisela Kastner, personal communication, August 08, 2003). Such an occurrence cannot be anticipated even if gross demand of a large number of past periods is available because the

148

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

change is abrupt, caused by an adjustment to the production plan (Gisela Kastner, personal communication, August 08, 2003). Based on this line of thought, it can be concluded that although exchanging gross demand might be beneficial in general, it is possibly only of very limited use in the engine supply chain. The reason for this is that the changes of net demand for a particular week quoted in different supply call-offs happen suddenly and randomly which makes net demand impossible to anticipate based on past gross or net demand data. In the future, however, demand in the engine supply chain might show a pattern, and then gross demand might be useful for determining future net demand. However, if it were not for these sudden net demand changes, no member of the supply chain would even have to try to anticipate future customer demand because the supply call-offs show demand for the next six months. Therefore, it has to be concluded that this scenario is only of questionable use in the engine supply chain. Constellation 2: Bilateral or Supply Chain-wide Exchange of Gross Demand with No Optimization If Constellation 2 is implemented, gross demand is the quantity which the supplier has to deliver to its customer. In this constellation, receiving gross demand is not extra, but binding demand information. Provided that gross demand changes as often as net demand does today (i.e. almost every week), the end result would be the same and there would be no effect on the extent of the bullwhip effect. Overall performance of the supply chain and its members might even worsen because there is no optimization of production, transportation, etc., either at the local or the global level. Constellation 3: Supply Chain-wide Exchange of Gross Demand with Supply Chain-wide Optimization of Order Quantities It was concluded in Constellation 1 that exchanging gross demand is possibly beneficial in general. Exchanging gross demand might even be more advantageous in terms of reducing the bullwhip effect if all members give up striving for their individual local optimum, and instead attempt to attain the supply chain-wide optimum (i.e. the global optimum). This supply chainwide optimum is achieved if transportation, production, etc., are optimized supply chain-wide. This is calculated by a central instance (e.g. a software tool used by the entire supply chain) based on gross demand along with inventory levels and capacity utilization which is made available by all supply chain members. Hence, if there is a central organization calculating order quantities (i.e. the net demand) of each supply chain member so that the supply chain-wide optimum is achieved, this constellation constitutes a form of central cooperation as described by Wyner and

4.3 Evaluation of the Audi AG Supply Chain

149

Malone (1996) and introduced in Section 2.3.4. In this section, it was noted that centralized cooperation leads to the lowest bullwhip effect compared to the other two cases (no cooperation and decentralized cooperation). Therefore, it can be stated that Constellation 3 might be very beneficial in terms of reducing the bullwhip effect (under the condition that variation in demand upstream in the supply chain is reduced further than downstream) and its negative consequences for the supply chain as a whole as well as for each individual company. Although implementing this constellation might be beneficial, problems might be encountered when trying to implement this scenario. First, one has to take into account that the complexity of centrally managing a supply chain might seem almost impossible to handle. However, if only certain aspects of a supply chain (e.g. the calculation of net demand for each echelon that is supply chain-optimal), and only a critical path are centrally managed, complexity might be reduced to a manageable level so that the benefits of central cooperation could possibly be enjoyed (e.g. a lower bullwhip effect). Another challenge that needs to be confronted when trying to implement this constellation is the fact that the members of a supply chain will probably be reluctant to give up their sovereignty and allow others to plan on their behalf. Closely related to this challenge is another challenge: Even if striving for the supply chain-wide optimum is more advantageous in terms of reducing the bullwhip effect and its consequences, it might, nevertheless, be difficult to convince supply chain members to forego local optimization and make gross demand and other relevant information available for calculating supply chainoptimal net demand at each level. The reason for this is that the position of some members might actually deteriorate if the supply chain-wide optimum is pursued. This is the case if companies are, for example, no longer able to produce in locally optimal lot sizes; it could result in higher costs for each company which that company would have to bear alone. Therefore, individual companies might be worse off than under the current system; even though supply chain performance as a whole increased. However, in the long run, the picture might be different. In the long run, the individual company is probably also better off – especially, if the members of the supply chain share the benefits gained by implementing this constellation fairly. The reluctance of the supply chain members to participate in supply chain-wide optimization can probably only be overcome if they understand this and also consider the long-term effects of their actions.

150

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

The described situation corresponds to the so-called “Repeated Prisoner’s Dilemma”105, that is, a Prisoner’s Dilemma with an infinite amount of rounds to be played (i.e. an infinite number of demand planning periods). According to Myerson (2001), in a Prisoner’s Dilemma with a finite number of rounds both parties will defect (i.e. optimize locally) because from their individual point of view this is the optimal strategy. Their losses are minimized if the other player defects, and it maximizes their payoffs if the other player cooperates (i.e. pursues supply chain-wide optimization; see Appendix A.2). Yet, if both players choose the non-cooperative or defective strategy, this leads to the worst possible outcome from the system’s perspective. As Figure 72 also shows, it would be best for the entire system if both players cooperated; however, this will not happen because defecting is the rational decision from the perspective of the individual player. In a Prisoner’s Dilemma with an infinite number of rounds to be played, the situation is the opposite. In such game, which resembles the situation in a supply chain, defecting is irrational and cooperating is rational both from the system point of view as well as from the point of view of the individual player because the system and each player maximize their payoffs if the cooperative strategy is followed in all rounds. This kind of game can be regarded as selfregulating which means that in such game, the optimal outcome from the system’s perspective is achieved automatically. The reason for this is that in such game, the players’ incentive to cooperate is “the hope of inducing future [cooperative] behavior by the other player” (Myerson, 2001, p. 309). Defecting might only generate a short-term advantage; but in the long run, the defecting player is off worse than if it had cooperated all the time because both will defect for the rest of the time after one player started to follow the strategy to minimize losses. The same line of argumentation can also be applied to a game with more than two players, e.g. the engine supply chain with four members. The only requirement for making this mechanism work in a game with two or more players is that all players can evaluate what the other players do. Software tools which facilitate the supply chain-wide exchange of information might help to provide sufficient information for such a situation because they might serve as a medium through which supply chain-optimal net demand

105

For a more detailed description of the Repeated Prisoner’s Dilemma and of the line of argumentation of the subsequent paragraphs, see Appendix A.2. The Prisoner’s Dilemma is discussed within the game theory and is a “theory of rational behavior for interaction decision problems. In a game, several agents strive to maximize their (expected) utility index by choosing particular courses of action, and each agent’s final utility payoffs depend on the profile of courses of action chosen by all agents. The interactive situation, specified by the set of participants, the possible courses of action of each agent, and the set of all possible utility payoffs, is called a game; the agents ‘playing’ a game are called players” (Sela & Vleugels, 1997). The Prisoner’s Dilemma is a game that deals with “the interaction and learning processes in populations of boundedly-rational agents” (Hoffman, 2001, p. 101).

4.3 Evaluation of the Audi AG Supply Chain

151

quantities at each echelon can be communicated. For example, if it is globally optimal for Audi Hungaria to demand fifty water pumps, this could be communicated to all companies via a supporting software tool. If Audi Hungaria ordered 150 water pumps from TCG Systemtechnik (because this is locally optimal), then all other companies in this supply chain would have the chance to see that Audi Hungaria does something that might be better for Audi Hungaria (i.e. it optimizes locally), but that might harm the supply chain as a whole as well as each individual member in addition to the culprit. No player will dare to defect if there is visibility because they know the consequences of this move. If one player defects, the others will start to defect in the following period; and they will all end up defecting for the rest of the game. This leaves each player with a worse payoff than it would have gotten if all players had cooperated in all periods. Based on the discussion in the above paragraphs, one can conclude that the supply chain-wide exchange of gross demand might be beneficial in terms of reducing the bullwhip effect if order quantities at each echelon are optimized not locally but globally. However, the bullwhip effect is only reduced under the condition that variation in demand upstream in the supply chain is reduced to a greater extent than downstream. Furthermore, it has been demonstrated at length by using the framework of the Repeated Prisoner’s Dilemma that cooperative behavior (i.e. ordering supply chain-optimal amounts) is assured automatically in such a system as long as there is an infinite number of planning rounds and as long as there is sufficient information in order to evaluate if a company cooperates or defects. In the engine supply chain, implementing this constellation might be beneficial in terms of reducing the bullwhip effect, however, only under one additional prerequisite. If no member is allowed to adjust its demand as it likes until the final supply call-off. If this were allowed, then exchanging gross demand and calculating supply chain-optimal net demand at each echelon would be of questionable use. The reason for this is that demand at a certain point in time, which a supplier has to satisfy, could still change from supply call-off to supply call-off; and the supplier would still have to produce an order based on demand information for this point in time given in earlier supply call-offs. Therefore, when it comes to demand uncertainty, the situation would be identical with the current situation. The only difference might be that not every party strives for its local optimum, but that the supply chain-optimum is aimed for; and this might lead to a better overall result. To summarize the discussion of this section, one can state that the benefit the use of supporting software tools like ICON-SCC, is probably questionable regarding the exchange of capacity utilization and inventory related information if each kind of information is considered separately. However, when it comes to exchanging gross demand along with information on capacity

152

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

utilization and inventory levels, this scenario might be beneficial in terms of reducing the variation in demand and, thus, possibly the bullwhip effect if order quantities are either optimized at the global or at the local level and if demand follows a certain pattern over time. Two further requirements for the constellation with global optimization are that there needs to be an infinite number of demand planning rounds, and that there be sufficient information to evaluate if the other players defect or cooperate. For the engine supply chain, a slightly differing conclusion has to be drawn concerning the advantages of exchanging gross demand information. If the supply call-offs contained reliable future (net) demand information, i.e. if demand quantities quoted in a supply call-off were fixed at a certain amount of weeks before the part has to be at the customer, an exchange of gross demand, as suggested in Constellation 1, would be unnecessary because there would be no demand uncertainty at all – no company would have to anticipate future net demand of its customer based on gross demand. Regarding Constellation 3, it has to be stated that the true benefit of this constellation can only be enjoyed if order quantities are fixed at a certain number of weeks before the ordered quantities need to be delivered. If this requirement is not met, demand uncertainty is not reduced, and the situation remains the same as today.

4.4 Summary of Results This case study uncovered an important bullwhip effect in the Audi A8 V8 4.0l diesel engine supply chain (see Table 19). As a conclusion for the reasons of this bullwhip effect in the engine supply chain, one could state that none of the four major causes of the bullwhip effect that Lee et al. (1997b) identified applies in this supply chain (see Section 4.3.1). The discussion in the section on “order batching” revealed that other factors probably cause variation in demand in the engine supply chain. However, the available information does not allow one to draw any definite conclusions for the causes of the bullwhip effect in this supply chain. Irrational order behavior as well as demand changes by Audi Neckarsulm, and probably also by Audi Hungaria, might contribute to this effect in the engine supply chain. The consequences of the bullwhip effect in the engine supply chain are difficult to identify because problems like over-supplies can have several roots. However, it is possible that the (temporary) over-supplies of thermostats at Wahler and TCG Systemtechnik as well as of water pumps at TCG Systemtechnik and Audi Hungaria (see also ICON 2003d) are the result of the bullwhip effect. Another reason could be, however, that the members of this supply chain maintain over-supplies as a safety buffer. Such over-supplies are costly in terms of inventory carrying costs. The other problems in the chain, e.g. correctional measures like extra shifts or

4.4 Summary of Results

153

expedited shipments, can hardly be contributed to the bullwhip effect. Instead, they seem to be the consequence of bottlenecks due to quality problems and the limited production capacity of companies not included in the critical path of the analyzed supply chain (Katalin Szorady, personal communication, July 16, 2003; Markus Bedic, personal communication, July 04, 2003; Thomas Hanin, personal communication, July 10, 2003). However, it was identified that, although participants of this supply chain do exchange demand information six months in advance, they do not stick to the announced demand data to a large degree. For example, Audi Neckarsulm does not stick to its announcements; even though the production plan is fixed 4 weeks in advance (see Figure 39). This highly contradictory fact is a source of uncertainty for Audi Hungaria and might be one of the major causes for the increase in demand variability between these two echelons. In addition, Audi Hungaria has to deal with its level of commitment to the announced supply call-offs with TCG Systemtechnik. As already mentioned, if quantities are increased or decreased by more than 15 percent in the last week from one supply call-off to the next, Audi Hungaria has to cover all additional costs (e.g. costs that arise because of extra shifts or expedited shipping) (Katalin Szorady, personal communication, July 16, 2003). To protect itself from additional expenses in this regard, it seems plausible to assume that Audi Hungaria is ordering amounts that exceed its net demand as a protection against sudden demand increases by Audi Neckarsulm of more than 15 percent. In analogy, the level of commitment of TCG Systemtechnik towards Wahler is higher than the commitment of Audi Hungaria towards TCG Systemtechnik. In this relationship, TCG Systemtechnik must stick to the last four weeks announced in the supply call-offs. These demand amounts cannot be changed anymore which means that TCG Systemtechnik must anticipate four weeks of demand from Audi Hungaria with the only guideline that one of these weeks is likely not to change by more than ± 15 percent. Again, it can be assumed that TCG Systemtechnik increases its net demand in order to be able to react to sudden demand increases by Audi Hungaria. In the opinion of the author, this is the most remarkable finding of this case study in addition to the fact that such an immense bullwhip effect exists in the critical path of this supply chain. For a company, the level of commitment to its supply call-offs towards its suppliers is higher than the level of commitment of its customers which is an additional source of uncertainty that causes the bullwhip effect to increase (Chen et al., 1999, p. 434). This practice of distributing rolling call-off plans to suppliers announcing demands in advance, and agreeing to certain levels of commitment for current call-off orders is very common in the European automotive industry. The goal is to reduce uncertainty by giving suppliers a reference

154

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

to base their forecasts on. The result of reducing uncertainty directly affects the extent of the bullwhip effect in the supply chain (Chen et al., 1999, p. 434). Though, as shown above, an opposite effect arises when the levels of commitment to the rolling call-off plans between echelons do not correspond and are not synchronized. Then, uncertainty dominates the decisions regarding order amounts and seems to cause an accentuated bullwhip effect. As stated by Senge (1990), the environment for the bullwhip effect is provided by a lack of “system thinking” which clearly applies to the asynchrony of commitment on different echelons in the supply chain. This concurs with Forrester’s (1961) explanation of the system structure causing the amplification or suppression of changes in demand, among other reasons, because of decision-making mechanisms in the organization. It has to be proven by corresponding simulations if the negative effect of a higher commitment towards a company’s suppliers than from customers towards the company exceeds the positive effects of the exchange of demand information in advance. If that should be the case, rolling calloff plans as they exist today have to be rethought in the European automotive industry. In any case, it can be stated that a higher commitment towards the supplier results in a lack of planning flexibility and, thus, a lack of reaction scope toward the uncertainty of customer’s demand. Several possible benefits of reducing the bullwhip effect in a supply chain, particularly in the engine chain, are conceivable (based on the consequences of the bullwhip effect described in Section 2.2.4.2). First, it might be possible to properly adjust inventory levels, i.e. to keep the right amount in stock, instead of either too much or too little inventory both of which are costly. In Section 4.2.2, it was demonstrated that in some parts of the engine chain, the inventory level is high at times and low at others which might possibly be due to the bullwhip effect. For other parts of the chain, this information was not available. Second, it might be possible to adjust production capacities to a level reflecting actual end customer demand, i.e. to maintain neither too little nor too much production capacity. Especially upstream, production capacities in a supply chain, including the engine supply chain, are probably too large because demand in this part of the supply chain does not mirror all actual end customer demands due to the distortion of demand information. Demand in the upstream parts of the supply chain is much larger than further downstream as can be seen in Table 18. If only those amounts were produced that reflect end customer demand, production quantities would be smaller and, thus, less production capacity is needed. This would probably result in substantial savings.

4.4 Summary of Results

155

Third, it might be possible to reduce the number of correctional measures (e.g. expedited shipments or extra shifts) and their respective costs. It has to be stressed, however, that in the engine supply chain correctional measures do not seem to be a problem – at least not in the monitored critical path of this supply chain. The benefits of a reduced bullwhip effect might be reaped if the effect itself is reduced through cooperation (Chen et al., 1999, p. 435; Chen et al. refer to “strategic partnerships”). The evaluation of the cooperation scenarios has brought up the following results concerning applicability and advantages regarding the bullwhip effect in the engine supply chain. Table 20: Summary of Evaluation of Cooperation Scenarios106 Cooperation Scenario

I

II

III

IV

V

VI

VII

VII

Generally advantageous?

+

+

+

+

+

+

–

+

Implemented in the engine supply chain between… Audi Neckarsulm & Audi Hungaria

+

–

–

–

–

–

–

–

Audi Hungaria &TCG Systemtechnik

+

+

–

–

–

–

–

–

TCG Systemtechnik &Wahler

+

–

–

–

–

–

–

–

If implemented, does it help to keep the bullwhip effect at a lower level than it would otherwise be? Audi Neckarsulm & Audi Hungaria

+

n/a

n/a

n/a

n/a

n/a

n/a

n/a

Audi Hungaria &TCG Systemtechnik

+

+

n/a

n/a

n/a

n/a

n/a

n/a

TCG Systemtechnik &Wahler

+

n/a

n/a

n/a

n/a

n/a

n/a

n/a

If not implemented, can it be implemented and beneficial in terms of reducing the bullwhip effect? Audi Neckarsulm & Audi Hungaria

n/a

–

–

–

–

–

–

–

Audi Hungaria &TCG Systemtechnik

n/a

n/a

–

–

–

–

–

–

TCG Systemtechnik &Wahler

n/a

–

–

–

–

–

–

–

One of the aspects that has to be considered in more detail is the evaluation of Scenario I (see Section 4.3.2.1). It was mentioned that the members of this supply chain have engaged in logistics alliances with LSPs which helps the members keep their lot sizes smaller than if they were

106

Note: + means thatthis scenario might help to reduce the variation in demand of companies in a supply chain. If variation in demand upstream in the supply chain is lowered to a greater extent than downstream, the bullwhip effect is reduced. – means that this scenario does not bear any potential for reducing the bullwhip effect.

156

4. The Supply Chain of the Audi A8 V8 4.0l Diesel Engine – A Case Study of Audi AG

carrying the freight on their own. This is so because if they were to insource transportation, they would not be able to consolidate their freight with freight from other companies in order to achieve a large volume to be transported. However, the design of these alliances results in overdimensioned lead times between the members of this supply chain. This is clearly a cause for an increase in the bullwhip effect (Lee et al., 1997b; Chen et al., 1999, p. 421). Table 21 shows an overview of current (LSP driven) and potential (if direct transportation was implemented) lead times in this supply chain. Table 21: Lead Times between Members of the Engine Supply Chain Current Lead Time

Lead Time by Direct Transportation

From Wahler to TCG Systemtechnik

3 Days

1 Day

From TCG Systemtechnik to Audi Hungaria

4 Days

6 Hours

From Audi Hungaria to Audi Neckarsulm

2 Days

1 Day

Relationship

It becomes apparent that due to the incursions in a logistics alliance with corresponding LSPs, the members of this supply chain face two effects that influence the extent of the bullwhip effect. On the one hand, this scenario reduces lot sizes and, thus, lowers the bullwhip effect. On the other hand, the consolidation of freight and the routing of this freight through specific hubs and spokes results in higher lead times than if transportation was implemented through direct truck shipping. In order to evaluate which effect is being affected in this supply chain, corresponding simulations have to be conducted in the future. These simulations are to provide an orientation on when direct shipping is the more effective alternative than the decision to participate in a logistics alliance with an LSP. Another simulation that could reveal improvements in the bullwhip effect is the question of implementing sequenced delivery between Audi Hungaria and Audi Neckarsulm (see Section 4.3.2.8). Two characteristics affect the bullwhip effect in opposite directions. First, the engines are shipped in frames that contain 6 equal engines and delivered first by truck and then by train to Audi Neckarsulm. This standardized frame contributes to lower the lead time in this multishipping transport and, thus, the bullwhip effect. Second, an implementation of sequenced delivery would lower the time span for assembling this part which would also lower the overall lead time and, thus, contribute to lowering the bullwhip effect. The problem here is that the different variants of the engine do not fit in the standardized frame which means that frame transportation would have to be given up. This would make the loading from truck to train more

4.4 Summary of Results

157

difficult and would result in longer handling times and, thus, contribute to increasing the bullwhip effect. It is a task for future research efforts to determine how these scenarios affect the overall lead times and, thus, the bullwhip effect in the context of corresponding simulations. In addition, it has to be considered if VMI might be the better strategy in the engine supply chain between Audi Neckarsulm and Audi Hungaria as well as between Audi Hungaria and TCG Systemtechnik. Since advance information on demand amounts does not always seem to be very reliable, VMI would probably contribute to reducing demand uncertainty and, thus, lower the bullwhip effect (Oliver, 1999; Hall, 2002; Waller et al., 2002). It is, though, questionable to what degree the reduction would be, and if the costs of additional means of storage would be lower than the savings through this pull strategy.

5 SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

There is much research work on the value of information in supply chains. For instance, Cachon and Fisher (2000) show in a numerical study that in the context of inventory management a full information policy between the actors of a supply chain can lower the total costs by 2.2% on average compared to a traditional non-sharing information policy (Cachon & Fisher 2000, Figure 1). They also show that far bigger benefits are acquired by other effects of implementing information technology: lead times reduction and smaller batch sizes. When cutting lead times in half, the supply chain reduces its total costs by 21% on average while when cutting batch sizes in half the costs are reduced by 22% on average (Cachon & Fisher 2000, Table 1). Gavirneni et al. (1999) estimate the savings due to information flow, and analyze when information is most beneficial. They focus their attention on the relationship between information, capacity, and inventory from the point of view of a supplier dealing with one customer (retailer), and compare a traditional non-information sharing policy with partial and complete information sharing policies. They show that information is most beneficial when the supplier has a high capacity and the demand variance of ƅ = S – s is moderate. In addition, Lee et al. (2000) address the exchange of demand information between a manufacturer and a retailer, and show that the manufacturer can obtain inventory and cost reductions with information sharing; especially if the demand is highly correlated, highly variable, or when the lead times are long. The planning with relevant information shared by actors of the supply chain represents one of three alternative cooperation forms as described by Wyner and Malone (1996) which is decentralized planning with information exchange. As mentioned in Chapter 2, this dissertation goes a step farther and also contemplates the shift of decision power in the planning process towards centralized decision making. This section shows an application prototype for evaluating the three alternative planning strategies:

160

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains •

Centralized, cooperative planning: referring to actors in the supply chain who plan together centrally as if they were one (virtual) company,

•

decentralized, cooperative planning: referring to actors who plan individually but exchange information that is relevant to the planning procedure of each company, and

•

decentralized, non-cooperative planning: referring to actors who plan independently and there is no information sharing at all for planning purposes.

One of the goals of the implementation of this application prototype is to serve as proof-ofconcept for the use of XML and Java to ensure maximum flexibility in both the implementation process and the scalability and modularity of the application. The architecture that underlies the prototype presented in this chapter uses state-of-the-art technology, mainly from the field of Java and XML, and uses architectural patterns that ensure a separation of user interface, logic, and data storage and analysis. This is achieved by decoupling the modeling of planning scenarios from the planning method for solving the problem. XML interfaces assume the communication between the components of the architecture, and Java provides the required platform independence. The application processes are designed for the one goal of comparing the results of different cooperation strategies on a single planning scenario in order to ensure comparability. The difference between the three planning strategies mentioned above offers the possibility to aid the strategic decision process of cooperation with business partners. For example, the difference in total costs of the supply chain for transportation activities between centralized and non-cooperative planning may be interpreted as the upper bound value of a system to support centralized transportation planning (e.g. a supply-chain-wide SCM software solution). In analogy, the difference between non-cooperative planning and decentralized cooperative planning in which actors exchange relevant planning information might be interpreted as the maximum costs of a supply-chain-wide implementation of, for example, an EDI standard for supporting automated information exchange. The goal of the architecture presented here is, thus, to address what seems to be the main obstacle for extending cooperative behavior in supply chains of the European automotive industry: The quantification of the benefits of cooperation (see Section 3.2). This chapter starts with the description of the prototypic implementation of the SCOptimizer (Section 5.1). First the architecture that is the basis for all processes of the application is presented (Section 5.1.1) and then the process of evaluation is described if cooperative planning uses the examples of transportation (5.1.2.1) and the bullwhip effect (5.1.3). The second part of

5.1 Prototypical Implementation

161

this chapter (Section 5.2) presents a computational study that shows how this prototype can aid the decision making process of which cooperative strategy to follow with business partners. Using a distribution scenario as an example, this numerical study provides a systematic method to evaluate the added-value of different degrees of cooperation using the SCOptimizer.

5.1 Prototypical Implementation107 The SCOptimizer is part of an application system called SIMPLEX (Supply Chain Management Platform Enabled by XML; Buxmann et al., 2002). While the other modules of SIMPLEX support the operational aspects of SCM, like the exchange of business documents between the partners in supply chains and the transformation between diverse XML business vocabularies, the SCOptimizer covers the planning side of SCM, in which it addresses the quantification of different degrees of cooperative planning.

5.1.1 The SCOptimizer Architecture The SCOptimizer is mainly based upon open standards, open source software, and freeware. It is written in Java and uses XML for the description of interfaces and the modeling of evaluation scenarios which makes the prototype platform independent.108 Java and XML ideally complement each other (Bosak, 1997). A wide range of XML components for Java-based applications are available. Free XML parsers and XSLT processors can easily be integrated into a Java application. In order to keep parser-independence, the prototype uses the JAXP (Java API for XML Processing) technology which decouples the application from the particular parser and processor used (McLaughlin, n.d.). The SCOptimizer uses the parser Xerces from Apache (http://xml.apache.org). For the graphical user interface, the Swing and AWT classes available in Sun’s JDK are used. The following figure shows the architecture of the SCOptimizer:

107

This prototypical implementation is an auxiliary means of research as described by König (1994) in his paper on the profile of “Wirtschaftsinformatik” (the German “Wirtschaftsinformatik” can be understood as the counterpart of the Anglo-Saxon “Information Systems” research field).

108

For the complete source code and the files involved in the implementation of the SCOptimizer prototype see Appendix C.5.

162

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Modeling of scenarios

XML

Optimization class

Description

XML Input data

XML Comparison of results

Optimization results

Graphic display

Figure 41: The Architecture of the SCOptimizer

The idea of decoupling the modeling of scenarios (here, referring to the description of actors and their relation to each other; see also Figure 50) from the particular use of optimization methods and planning models is realized in our prototype through XML. This technology acts as the mediator between a particular scenario (“Modeling of scenarios” in Figure 41) and the available planning methods (“Optimization class” in Figure 41). After modeling a scenario, the planner selects the planning task that is to be performed. At this stage, the SCOptimizer looks for all available task-specific planning models and dynamically displays a list of the models so that the user can select the appropriate model and method for evaluation. Every planning model and every method are described in an XML file (“Description” in Figure 41), which is stored in the file system (in a later version of the application prototype, the XML data base Xindice will serve as storage container for better access performance; see http://xml.apache.org/xindice). This description is read at runtime by the prototype and is used to dynamically create the input masks and to instantiate the optimization class. This description allows the SCOptimizer to create the appropriate input masks at runtime so that the user can enter the required data for the planning and optimization methods (see Figure 53). This data is also stored in an XML file (“Input data” in Figure 41), which is then parsed by the optimization class. The class applies the optimization or heuristic method and stores the result in another XML file (“Optimization results” in Figure 41; see also Figure 47). The patterns for optimization classes in the context of our prototype call for a graphical display of the results (“Graphic display” in Figure 41). The graphic display as well as the XML results can be used for a comparison between the different degrees of cooperative planning (“Comparison of results” in Figure 41).

5.1 Prototypical Implementation

163



Clarke and Wright - Savings algorithm. A classic algorithm from 1964 (Clarke and Wright) to solve Vehicle Routing Problems with capacity restrictions. There is no restriction in the number of vehicles and there is only one depot. The algorithm starts creating routes from the depot to every node and back. At each step, the two routes that realize the largest costs savings are merged.

true true true true true

Costs per distance-unit for transporting one product item.

all

all

SourceName DestinationName

...

...

Figure 42: Excerpt from an XML Description for a Solver Class

All needed and created XML documents comply with a strictly defined structure and vocabulary. This avoids the hard-coding of the offered planning models and optimization methods in the evaluation prototype. These models and methods are accessed at runtime and thus can be added to and removed from the prototype without additional programming effort. The optimization classes of course have to provide the appropriate XML interface. This design pattern allows a nearly unconstrained scalability of the prototype since the inclusion of further optimization and heuristic methods is performed by adding a reference in specific XML registries which are accessed by the prototype at runtime (see Figure 43). Due to the design patterns which include the XML interfaces of a planning class as well as its behavior the application is able to provide additional planning methods in the user interface masks and, if selected, create the required input masks at runtime and integrate the graphic display of results in the application framework.

164

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains



Clarke and Wright with PointGraph Clarke and Wright with Matrix Gillett and Miller with PointGraph Gillett and Miller with Matrix Best Of All with PointGraph Best Of All with Matrix Branch and Bound

The solution will not be improved. Remove 2 edges and add 2 other edges iteratively until no better solution can be achieved. Remove 3 edges and add 3 other edges iteratively until no better solution can be achieved.

Figure 43: XML Registry for at Runtime Available Distribution Planning Methods

The goal of keeping the application as flexible as possible turns out to be relatively easy with the use of XML for storing parameters and configuration options. For instance, the available types of actors that take part in a supply chain scenario and are available in the modeling masks are defined in corresponding XML files. This allows to specifically change the set of actors for any given scenario which in turn permits almost any possible context for planning even beyond the context of SCM.

Warehouse false warehouse ./images/warehouse.jpg ./images/medium_warehouse.jpg ./images/small_warehouse.jpg

Retailer false retailer ./images/dealer.jpg ./images/medium_dealer.jpg ./images/small_dealer.jpg

Transporter true supply ./images/supply.jpg ./images/medium_supply.jpg ./images/small_supply.jpg /

Figure 44: Excerpt from the XML Registry of Available Actors for Modeling

In addition, the specific characteristics of actor types available for modeling are described in corresponding XML files that carry the name described in the actor’s registry as shown in Figure 44. For instance, a warehouse will be declared in the file “warehouse.xml” as shown in Figure 45. This file fulfills several goals within the prototype. First, it declares the planning methods that can

5.1 Prototypical Implementation

165

be applied for this single actor (for example, lot sizing in the warehouse; see the tag “optimization” in Figure 45). It also describes which classes are used for performing this planning (see the attribute “className” in Figure 45) and it describes the needed input (see the tag “neededInput” and its sub-elements “InpParam” in Figure 45) as well as the data type of this input. This allows the application to dynamically create the input masks if the planner seeks to plan the optimal lot size of a single warehouse. On the other hand, it also declares the output as a result of applying the selected planning method (see the tag “resultOutput” and its sub-elements “output” in Figure 45). Furthermore, available types of input parameters and dimensions (e.g. currencies, weight, size, or volume dimensions) are also declared in corresponding XML files (see Appendix C.5). When planners create the model of a supply chain or its parts in order to perform the quantification of the different degrees of cooperative, the application creates an environment and stores all relevant data in a separate directory in the file system. For instance, the creation of a distribution scenario, as described in Section 5.1.2.1, results in the creation of a new so-called “desk session” and its corresponding XML description file. This file takes over the storage of the parameters of the model created by the planner listing all active actors in the model (see the tag “DeskNodes” and its sub-elements “Node” in Figure 46) and all of its transportation and information connections (see the tag “Connection” and its attribute “Type” in Figure 46). In a distribution scenario as described here, also relevant demand data (see the tag “demand” in Figure 46) as well as geographical coordinates (see the tags “x” and “y” in Figure 46) are included as well. Furthermore, the architectural patterns of the application provide the possibility of additional parameters for each actor which are stored in corresponding XML files (see the tag “DataFile” in Figure 46). For the planning context presented in this dissertation, no additional specific data were required that could not be provided by the standard input masks of the planning classes. Thus, no additional data files were implemented. As already mentioned, the application prototype creates an environment called “desk session” for each model. The “.desk” file and the corresponding directory (named exactly like the .desk file) provide the required structure for performing planning activities on the basis of the created model. All results of selected planning methods are stored in the corresponding session directory and are, thus, available for comparison and analysis. In the first place, all results are stored in an XML file called “optdata.xml” which is a temporary file that is overwritten by each planning class when applying the planning method to the model. The design patterns require, though, that planning classes permanently store the results of a planning run. When doing this, the planner is

166

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

able to provide a name for the results file which is then stored in the desk directory. Results files are stored with a specific prefix “r_” which allows the system to always be aware of available planning results for a certain model. This “awareness” allows the application to have access to the graphic display of planning methods without always having to perform planning runs. It also enables the system to refill input masks with data of existing planning runs which eases to a large degree the handling of multiple variations of planning scenarios.

Classical Economic Lot Size Model The Classical Economic Lot Size Model, introduced by Harris(1915) and Wilson(1934), faces a constant demand for single item and places orders for the item from another facility, possibly a warehouse or a retailer, in the distribution network wich is assumed to have unlimited quantity of the product. Demand is constant at a rate of D items per unit time. Order quantities are fixed at Q items per order. A fixed set-up cost K is incurred every time the warehouse places an order. A linear inventory carrying cost h is accrued for every unit held in inventory per unit time.



Price per unit The price of a single unit of the corresponding item.

Dimension in which the transported items are measured dimensions.xml

…



Optimal cycle time The optimal time between two successive replenishments.

getCycleTime

…

…

Figure 45: Excerpt from the Description File for Warehouses

5.1 Prototypical Implementation

167



3rd Party LSP supply 500 20 3rd Party LSP.xml ./images/small_supply.jpg ./images/medium_supply.jpg ./images/supply.jpg 0 0 0



DC 1 harbor 140 140 DC 1.xml ./images/small_harbor.jpg ./images/medium_harbor.jpg ./images/harbor.jpg 30 123 0





…

Figure 46: Excerpt from the XML Description of a Desk Session

As shown in Figure 47, a results file includes all relevant input data for a planning scenario (e.g. “vehicleCostPerTour” and “Tab” tags in Figure 47) as well as all relevant output parameters that lead to the final result of a specific planning activity and allow corresponding analysis and graphic or tabular display (see “central” tag and its sub-elements “createdTour” in Figure 47). With this premise, the application is able to provide the environment for creating models of supply chains that are independent of the methods a planner can apply to them. This allows the possibility of modeling a supply chain and performing several planning runs from different areas (e.g. distribution planning and evaluation of the bullwhip effect) without having to change the initial supply chain model. If developers stick to the design patterns described in this section, the planning alternatives can be extended at runtime without interfering with the planner’s activities. This flexibility is acquired through the use of XML and its seamless integration within the Java programming language. The process side of the application, referring to how the planner can create supply chain models and how different degrees of cooperative planning can be applied, is described in the next two sections. The author has implemented two planning areas that are of great importance in SCM and provide a context for quantification of the advantages of cooperation. First, Section 5.1.2 describes the evaluation process in the context of distribution planning where several retailers are

168

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

to be supplied with a certain good from a set of distribution centers. Here, the vehicle routing is to be planned. Second, Section 5.1.3 describes the process of evaluating the advantages of cooperation in a simple supply chain in the context of the bullwhip effect.



2 4 3

0



1 central 500

229 169 231 162 237 177 350 323 179 239 388 316

...



283.4556916317886 408.0 283.45569163178857

291.45569163178857 2: DC 1



74.32361670424818 323.0 74.32361670424818

82.32361670424818 2: DC 1

...

Figure 47: Excerpt from a Results File for Centralized Distribution Planning

5.1 Prototypical Implementation

169

5.1.2 Prototypical Implementation of the Evaluation of Cooperative Distribution Planning with the SCOptimizer 5.1.2.1 Planning Background A very common planning activity in supply chains is transportation planning. This occurs as a part of the many logistic activities in supply chains. A typical example for the need of transportation plans can be found in the context of distribution planning (Bramel & Simchi-Levi, 1997, pp. 67ff). In this case, a number of distributors delivers goods to a number of customers in order to satisfy their demand for those goods within a given time frame.109 In a non-cooperative context, each distributor can plan the delivery, for example, in accordance with the model known as Capacitated Vehicle Routing Problem (CVRP) (Dantzig & Ramser, 1959; Balinski & Quandt, 1964; Domschke, 1997, pp. 211f): n

Min

n

F ( x) = ¦¦ cij xij

(5.1)

i =0 j =0

With following constraints: n

¦x j =0

ij

=1

for i = 1,…,n

(5.2)

ij

=1

for j = 1,…,n

(5.3)

n

¦x i =0

¦¦x i∈Q j∈Q

ij

xij ∈ {0,1}

≤ Q − r (Q) for any subset Q ⊆ V − {0} with Q ≥ 2

for i,j = 1,…,n

(5.4)

(5.5)

This model is characterized by a linear objective function and a one-level structure. It describes a one product case in which every customer (who is part of the network G with a total of V =

109

For a comprehensive introduction and overview to vehicle routing and scheduling problems refer to Domschke (1997, pp. 204ff).

170

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

0,1,…,n nodes) demands the same product but in different quantities bi. This demand has to be satisfied by a single distributor (node 0) who provides enough available stock of this product. This distributor holds a vehicle fleet that is large enough and whose trucks cover a predetermined maximum capacity cap. The covered route between two nodes (i and j, with i ≠ j) causes specific costs cij. The goal of this delivery problem is to minimize the total cost of transportation110 which is equivalent to the minimization of the number of routes (i.e. the number of required vehicles) and the total covered distance (see Equation (5.1)). The Constraints (5.2) and (5.3) guarantee that only one vehicle leaves from and arrives at each node. Constraint (5.4) ensures that the vehicle capacity is not exceeded and that every route includes the distribution center (node 0). In this constraint, r(Q) represents the minimum amount of required vehicles, for example: ª b º r (Q) = «¦ i » cap « i∈Q »

(5.6)

Because of the assumption that splitting of delivery orders is not allowed, the following constraint applies:

bi ≤ cap

for i = 1,…,n

(5.7)

In the context of comparing the three different cooperation strategies for a scenario in a supply chain with three distribution centers, each of which supplies four customers, the non-cooperative delivery planning would mean that each center solves a CVRP that only considers those customers with whom this distributor has direct business relationships. Other customers supplied by other distributors of the same supply chain are not taken into account. The concept of this scenario can result in a setting as shown in Figure 48: It conceptually shows that the distribution centers (see the boxes in Figure 48) sometimes deliver customers that are located nearer to another distribution center (see the circled customers in Figure 48; the lines represent the closed tour of a vehicle). This means that the total covered in the supply chain distance could probably be minimized if a distributor would supply those customers nearest it (assuming this does not result in the need for additional vehicles).

110

For a detailed am overview on transportation costs characteristics refer for example to Ballou (1999, pp. 153ff) or Bowersox and Closs (1996, pp. 368ff).

5.1 Prototypical Implementation

171

Figure 48: Non-Cooperative Transportation Planning

In a centralized cooperative setting, distributors will underlie the “team assumption” which means they will prioritize the overall result above the individual results (see Section 2.3.2). This allows giving up the bonds of business relationships with customers and permitting a “customer exchange” between the cooperating distributors. In this case, customers would be assigned to the nearest warehouse and the delivery routes would be planned accordingly. This kind of model is called Multi Depot Vehicle Routing Problem (MDVRP; Golden et al., 1977; Laporte et al., 1988). One way of consecutively solving the assignment and the routing problem is applying the Voronoi heuristic first and then solving the CVRP for each distributor (Voronoi, 1908; Klein, 1989; for an alternative heuristic procedure see for example Chao et al., 1993). This method divides the planning area R2 into so-called Voronoi regions by using the following formula:

V ( DCi ) =

 {x ∈ R

k :k ≠ i

2

: d ( x, DCi ) < d ( x, DC k )}

(5.8)

The variable x represents the location of a customer that has to be supplied. This customer is assigned to the distribution center DCi that is nearest to it in terms of distance d. The result of applying this formula is a set of regions V; each supplied by a single distribution center DCi. Assuming that the distance is the primary cost driver and all distributors dispose of enough handling and transportation capacities, this reassignment will often lead to reduced overall tour costs. In analogy to Figure 48, the concept of this scenario could result in a setting as shown in Figure 49:

Figure 49: Centralized Cooperative Transportation Planning

172

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

While Figure 48 depicts the non-cooperative context where each warehouse plans the share of the trucks, load allocation, delivery order, and delivery route for their customers independently, Figure 49 conceptually displays the results of a centralized cooperative planning. Under the assumption that the distance-dependent variable costs have the highest proportional priority, cooperative planning will result in overall lower transportation costs for the supply chain (the lines in Figure 49 represent the closed tour of a vehicle). This scenario might sometimes result in single distributors performing worse (in terms of costs) than in the non-cooperative environment as seen above. If, for example, after the reassignment of customers, a distributor must supply more demand than before, this might cause the distributor to need more vehicles and maybe to supply more customers. In a centralized planning environment, the team assumption (see Section 2.3.2) will, though, make this result feasible. On the other hand, this will not work in a non-team based situation like in the decentralized cooperative scenario. In this case, the distributors do not give up decision power (as in the case of the reassignment of customers which is performed centrally), but rather try to take advantage of the planning results of cooperating distributors in form of exchanging information. If we assume that the fleet of vehicles at each node is limited and that there exists a 3rd party LSP that provides vehicles at a higher fixed rate than the fixed costs of self-owned vehicles, then the exchange of unneeded vehicles between cooperating distributors might be of advantage. For instance, in a non-cooperative environment a distributor DC1 needing additional vehicles to serve the demand will only be able to purchase transportation capacity from the LSP. In the decentralized cooperative scenario, though, if another distributor DC2 has unused vehicles and the costs of passing them on to DC1 are lower than the rate of the LSP, there will be a cost advantage for DC1 as well as for the supply chain. DC2 will remain in the same position since it receives the costs for the vehicle. In this scenario, neither of the collaborating distributors gives up any decision power nor does it underlie the team assumption. They collaborate in a “winsame” situation and also contribute to a better supply chain result. Each distributor performs a CVRP as in the non-cooperative environment. This time, however, there are several sources of transportation capacity with vehicle costs. This exchange of transportation capacity between partners in the supply chain can be put into practice in different settings and environments (Walter, 1975, p 25; Richardson, 1998). Since there is no central instance in this scenario, the vehicle exchange will take place in the context of a series of informational contacts between collaborating distributors and not as a result of centralized optimization that considers all relevant parameters. For instance, there will be contact rounds via telephone which will provide each actor with the required information. If,

5.1 Prototypical Implementation

173

for example, DC1 needs an additional vehicle, it will contact DC2 and DC3 asking for unneeded vehicles and the corresponding transfer costs. DC1 will then compare the costs of the transfer of these vehicles with the LSP rates and choose the less expensive source for the required vehicle. On the other hand, if DC2 has an unneeded vehicle in its fleet and both DC1 and DC3 require an additional vehicle, there will probably be a subjective preference (based, for example, on past relationship experiences with the other distributors) for DC2 that will influence the decision of whom to transfer the vehicle to. In the context of this dissertation, the assumption here is to dismiss any subjective factors and consider only objective measures that influence decisions. Thus, it is assumed that DC2 will consider the lowest transfer cost as the first option and the highest as the second. This influences the overall supply chain result and leads to the following heuristic for the exchange of vehicles between distributors and the LSP: •

Step 1: Divide the distributing nodes in the supply chain into two separate sets. One set contains all distributing nodes that, after solving their initial CVRP, have unused vehicles. The second set contains all distributing nodes that, after solving their CVRP, need additional vehicles. Distributing nodes that neither need additional vehicles nor have unused vehicles are not taken into further consideration.

•

Step 2: Define potential transfer relationships among distributing nodes that might enable a vehicle exchange. Depending on existing informational relationships (defined in the modeling user interface; see Figure 50), all possible transfer relationships between distributors that offer unneeded vehicles and distributors that require additional vehicles are declared and recorded.

•

Step 3: Assign each potential transfer relationship the maximum amount of vehicles that can be transferred thereby. The maximum amount of vehicles that can be exchanged through a transfer relationship consists of the minimum of two measures: The number of unused vehicles at the offering distributor and the number of additional vehicles needed by the demanding distributor.

•

Step 4: Assign to each potential transfer relationship the transfer costs per vehicle. Transfer costs consist of fixed costs for the vehicle use and variable costs. Fixed usage costs are distributor-dependent, and variable costs are distance-dependent.

•

Step 5: Create a ranking of all potential transfer relationships. The ranking is in accordance with the transfer costs per vehicle. Potential transfer relationships with the lowest transfer costs per vehicle will be ranked higher than relationships with higher transfer costs per vehicle.

174

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains •

Step 6: Exchange as many vehicles as possible through the less expensive transfer relationships. The maximum amount of vehicles in each relationship will be exchanged in accordance with the ranking. If either all vehicles at the offering node are exhausted or the demanding node does not require any additional vehicles, the next transfer relationship in the ranking is activated. Transfer relationships in the ranking will be activated until there are no more distributors left who still have a demand.

This heuristic aims to emulate the results of a decentralized coordination through a series of contacts between the collaborating parties. In analogy to a decentralized coordination in reality, although this course of action follows rational behavioral patterns, there is no guarantee that it will lead to the best result since there is no central entity that optimizes the problem (see Buxmann, 2001, pp. 120ff for a comparable scenario in the context of an investment problem).

5.1.2.2 Description from the Planner’s Point of View This section gives an overview on how the SCOptimizer can be applied from the planner’s perspective to determine the value of different cooperative degrees in the context of distribution planning. The first task for the simulation is the modeling of a supply chain with an arbitrary number of participants. Thus, the SCOptimizer is capable of planning a number of different levels of distribution; so, for example, one can simulate the supplying of manufacturers by suppliers as well as the provisioning of wholesalers by the manufacturers.111 In the following section, we have chosen a distribution scenario similar to the one mentioned in the preceding section (see also Figure 48 and Figure 49) in which a set of distribution centers (DCs) provide retailers with a certain product they demand and will perform a CVRP (or an MDVRP in the centralized planning alternative). The goal is to satisfy this demand at the lowest possible costs. These costs consist, on the one hand, of variable distance-dependent costs and fixed vehicle costs, on the other hand. Actors may either use vehicles from their own fleet, or they may rent vehicles from an external logistics service provider. Furthermore, in the cooperative scenarios actors with unused vehicles may provide other actors with missing transportation capacity through vehicles. Moreover, the team assumption applies in the centralized cooperative scenario, i.e. for distributors, the overall result of the supply chain is more important than the individual result (Marschak, 1954; Marschak, 1955). The additional assumptions for this simulation model are:

111

For the complete source code and the files involved in the implementation of the solving class for the distribution planning, see Appendix C.3. For all the files involved in the exemplary evaluation of distribution planning, see Appendix C.1.

5.1 Prototypical Implementation •

There is only one product to distribute,

•

the demand has to be satisfied,

•

orders cannot be split,

•

vehicles have a fixed capacity, and

•

DCs have a limited amount of vehicles available.

175

Figure 50 shows the structure of the setting for our simulation in the modeling mask of the SCOptimizer prototype. In this mask, the planner has the possibility of adding nodes to the supply chain per “drag&drop”. Available types of nodes can be seen in the tool bar above the working area. Currently, implemented types of actors are: Warehouse, retailer, transporter (meaning freight forwarder), supplier tier 1 and tier 2, customer (meaning end customer), distribution center, wholesaler, and factory (meaning manufacturing plant). The right side of the toolbar presents two types of connections: simple and bidirectional. When planners select a connection type, they are able to create two types of relationships: Information connections (red arrows in Figure 50) and transportation connections, or flow of goods (black arrows in Figure 50).

Figure 50: Modeling Mask of the SCOptimizer

As an example, Figure 50 shows how the planner created a model where three DCs (DC1, DC2, and DC3) supply four retailers each (Retailer 1 through Retailer 12). In addition, the DCs

176

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

maintain information relationships (shown as bidirectional arrows in Figure 50) which represent the potential exchange of information about transportation capacity. The external LSP (3rd Party LSP in Figure 5) also exchanges this type of information with the DCs. After creating the model, planners will have to choose one of the planning methods available in order to begin with the evaluation. In the menu bar of the modeling mask, planners can select “Logistics” in order to access the set of available planning methods. In the case of distribution planning, there is a set of exemplary classes implemented that apply traditional methods in order to perform the routing: •

Savings method (Clarke & Wright 1964; Domschke 1997, pp. 244ff)

•

Sweep-Algorithm (Gillet & Miller 1974; Domschke, 1997, pp. 236ff)

•

Branch and Bound (Little et al., 1963; Smith et al., 1977; Domschke, 1997, pp. 125ff; Domschke & Drexl, 2005, pp. 148ff)

Figure 51 shows the resulting mask for choosing between available distribution planning methods. This mask is created at runtime and applies the description data contained in the XML files of those available classes (see Figure 42).

Figure 51: Selecting a Planning Method for the Vehicle Routing in the Distribution Planning

The methods with a point graph apply a single distance while the methods with a matrix allow asymmetrical distances for both ways of each transportation relationship in the modeled scenario (DRA Systems, 2000). The “Best of all” methods apply all available heuristics combined with a 3opt improvement algorithm and take over the best result. After choosing the method (e.g. Savings method with a point graph), the prototype parses its XML description and at runtime creates the input masks for the user to enter the required

5.1 Prototypical Implementation

177

information. For instance, if the planner chooses the Savings method with a point graph, the system will request the entering of a first set of parameters: •

Improvement algorithm. The planner can choose to apply an improvement algorithm or not. The selection, for example, offers a 2-opt and a 3-opt improvement algorithm (Domschke, 1997, pp. 118ff).

•

Cooperation degree and cooperating nodes. As shown in Figure 52, planners can select between one of the three cooperation degrees and, if needed, the nodes involved in the cooperation.

•

Tour characteristics. Planners may declare tours to be closed or open. in the latter case, the vehicles are not required to return to the “headquarters” where the tour started.

•

Global costs. In this case, the planner may enter both a limit for the total distance costs of a single tour and the global costs per distance unit.

•

Vehicle type. Planners may choose between available vehicle types and their specific transportation capacity. It should be noted that since orders cannot be split, the planner has to select a vehicle type that allows the shipment of the largest customer order.

•

Fleet characteristics. At this stage, the planner has to enter the number of available vehicles at each distributing node and the fixed costs per vehicle use. In the case of a third party LSP, there is no limitation for the availability of vehicles.

Figure 52: Mask for Selecting Cooperation Degree and Cooperating Nodes

Figure 52 shows that the planner has selected a decentralized, cooperative planning in which all three distributors cooperate by exchanging information on both unused vehicles and additional vehicles needed. As mentioned above, the planner can also choose to perform a centralized, cooperative planning. This would mean that the MDVRP applies in which every supplied retailer is first assigned to the nearest warehouse (using the Voronoi method described above; see

178

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Formula (5.8)), and then the selected method (in this case Savings) is performed for each distributor. If planners want to evaluate the non-cooperative context, they will select just one warehouse in the mask shown in Figure 52. In this case, there will be no reassignment of customers, only the chosen optimization or heuristic method will be applied. In this scenario, the non-cooperative planning has to be done once for each distributing node. After entering the first set of parameters, the system creates the second input mask for the node specific data. Figure 53 shows the input mask for the Savings method with a point graph.

Figure 53: Input Mask for the Savings Method of Clarke and Wright with Point Graph

The example of Figure 53 shows the distance between the involved partners of the supply chain. By clicking the other tabs in the mask, the planner is able to enter demand data for each retailer, coordinates on the plane for each actor in the supply chain (on which basis distances are automatically calculated), and the handling and storage capacity of each distributing node. This mask offers for all input tables (except for “Distances” since it is calculated by applying the location coordinates) the possibility of creating random values. For this purpose, the planner is able to enter the mean value and standard deviation, and the system fills in the tables with the created random values based on an equal distribution. To serve as an example, this section is going to present the results of the planning for the scenario described above using the Savings method and the 3-opt algorithm. Tours are closed, there is no cost limitation per tour, and the variable, distance-dependent costs amount to one cost unit per distance unit. Vehicles have a capacity of 500 load units and, as already mentioned, DCs have a limited number of vehicles which cause distributor-dependent fixed costs and distance-dependent transfer costs:

5.1 Prototypical Implementation

179

Table 22: Vehicle Parameters Vehicle Owner

Available Vehicles

Fixed Vehicle Costs

Transfer Costs

DC1

4

8

Distance-dependent

DC2

1

9

Distance-dependent

DC3

1

10

Distance-dependent

3rd Party LSP

Sufficient

75

0

In the non-cooperative planning, DCs have only one source of additional transportation capacity available which is the external LSP. The third party LSP has enough vehicles available which can be retrieved at a given price (75 cost units in this example). In both decentralized and centralized cooperative planning, DCs have an additional possibility: A DC may provide other DCs with its non-used vehicles for a fixed price equal to its fixed vehicle costs. Additional distance-dependent costs are incurred when transferring the vehicle from one DC to another. The complete set of input parameters for this example can be seen in the following tables: Table 23: Further Input Parameters Node

Demand

Coordinates (x/y)

DC1

-

30/123

Sufficient

DC2

-

76/32

Sufficient

DC3

-

61/83

Sufficient

3rd

Party LSP

Handling/Storage Capacity

-

0/0 (irrelevant)

-

Ret. 1

244

100/3

Sufficient

Ret. 2

317

114/215

Sufficient

Ret. 3

165

246/157

Sufficient

Ret. 4

182

126/50

Sufficient

Ret. 5

399

22/45

Sufficient

Ret. 6

100

135/184

Sufficient

Ret. 7

130

2/239

Sufficient

Ret. 8

26

41/87

Sufficient

Ret. 9

217

81/217

Sufficient

Ret. 10

207

50/217

Sufficient

Ret. 11

99

203/74

Sufficient

Ret. 12

380

15/157

Sufficient

180

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Table 24: Distance between Nodes in the Plane From /To

DC1 DC2 DC3 Ret. 1

Ret. 2

DC1

-

101.9 50.61 138. 7 92

124.5 218.6 120. 8 6 60

78.41 121.4 119. 3 33

37.64 106.9 96.1 4 0

179.8 37.16 1

DC2

101. 97

-

186.9 211.0 53.1 0 1 4

55.54 163.0 219. 5 83

65.19 185.0 186. 7 82

133.7 139.0 6 9

DC3

50.6 1

53.16 -

142.2 199.2 72.9 4 5 0

54.45 125.2 166. 1 78

20.40 135.4 134. 8 45

142.2 87.13 8

Ret.1

138. 92

37.64 89.00 -

212.4 212.2 53.7 6 1 1

88.59 184.3 255. 5 54

102.6 214.8 219. 5 4 76

125.1 175.9 0 0

Ret.2

124. 58

186.9 142.2 212. 0 4 46

-

193.3 37.44 114. 0 54

147.3 33.06 64.0 5 3

166.7 114.7 4 4

Ret.3

218. 66

211.0 199.2 212. 1 5 21

144.1 8

250.4 114.2 257. 4 4 41

216.6 175.5 204. 2 7 98

93.48 231.0 0

Ret.4

120. 60

53.14 72.90 53.7 1

165.4 160.7 4 8

104.1 134.3 226. 2 0 05

92.70 172.9 183. 6 48

80.65 154.1 8

Ret.5

78.4 1

55.54 54.45 88.5 9

193.3 250.4 104. 0 4 12

-

46.10 181.8 174. 4 26

183.3 112.2 1 2

Ret.6

121. 43

163.0 125.2 184. 5 1 35

37.44 114.2 134. 4 30

179.1 4

135.0 63.29 91.1 7 8

129.3 123.0 2 0

Ret.7

119. 33

219.8 166.7 255. 3 8 54

114.5 257.4 226. 4 1 05

195.0 143.9 3 2

156.9 82.01 52.8 2 0

260.0 83.02 5

Ret.8

37.6 4

65.19 20.40 102. 65

147.3 216.6 92.7 5 2 0

46.10 135.0 156. 7 92

-

162.5 74.67 2

Ret.9

106. 94

185.0 135.4 214. 7 8 84

33.06 175.5 172. 7 96

181.8 63.29 82.0 4 1

136.0 1

Ret.10 96.1 0

186.8 134.4 219. 2 5 76

64.03 204.9 183. 8 48

174.2 91.18 52.8 6 0

130.3 31.00 1

209.4 69.46 2

Ret.11 179. 81

133.7 142.2 125. 6 8 10

166.7 93.48 80.6 4 5

183.3 129.3 260. 1 2 05

162.5 187.9 209. 2 7 42

-

Ret.12 37.1 6

139.0 87.13 175. 9 90

114.7 231.0 154. 4 0 18

112.2 123.0 83.0 2 0 2

74.67 89.20 69.4 6

205.5 1

53.16 37.6 4 89.0 0

Ret. 3

Ret. 4

144.1 165. 8 44 160. 78

Ret. 5

Ret. 6

Ret. 7

179.1 195. 4 03 143. 92

Ret. 8

Ret. 9

Ret. 10

136.0 130. 1 31 31.0 0

Ret. 11

Ret. 12

187.9 89.20 7

205.5 1

The next step after entering this data is to perform the planning run. For this purpose, the planner will press the “Create Plan” button (see Figure 53) and the input data will be serialized to the corresponding temporary XML file. After applying the selected planning method, the planning classes store the results in the same XML file and offer to store it permanently. After selecting the name of the permanent file, the planning classes open their analysis mask. In the case of distribution planning, the SCOptimizer offers both a tabular and a graphic representation of the resulting tours and an overview of the key parameters of the resulting routing.

5.1 Prototypical Implementation

181

For the example described above, the decentralized, non-cooperative planning leads to the following results.

Figure 54: Tabular Display of Results for the Decentralized, Non-Cooperative Planning

Figure 54 shows the results of the decentralized, non-cooperative planning in which all three DCs plan their tours independently. The upper part of the mask shows the tabular display of the resulting tours. In the center part, planners can see details on the tours; namely, the number of delivered nodes, tour distance, tour costs, and transported quantity. The bottom part with the white background shows an aggregated view of the resulting tour plan. This non-cooperative scenario results in overall supply chain costs of 2361.69 cost units. Each DC requires two tours to satisfy the demand of four retailers each. DC2 and DC3 require one additional vehicle each which they retrieve from the only source available, the external LSP. DC1 remains with two unused vehicles. In order to see the graphic representation of the results, the planner can press the “Show Graph” button.

182

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Figure 55: Graphic Display of Results for the Decentralized, Non-Cooperative Planning

Figure 55 shows the set of tours on the plane where the arrows indicate the vehicle movements. DC1 performs two tours (tours 6 and 7) and uses two of its four vehicles. DC2 performs tours 8 and 9 and uses its own vehicle and one vehicle from the external LSP (the transfer arrow is labeled with a “V” between the LSP and DC2). DC3 performs tours 10 and 11 and also retrieves an external vehicle from the LSP (also labeled with a “V”). As can be seen, some of the supplied retailers are delivered from a DC that is not the nearest to them. For instance, Rte. 12 is nearest to DC1 but is delivered from DC3, Rte. 7 is nearest to DC1 but is delivered from DC2, and Rte. 4 is nearest to DC2 but is delivered from DC 1. It is also noticeable that distributors do no take into account that other DCs might have unused vehicles that might be less expensive to acquire than those from the external LSP. As mentioned before, it is this information on unused vehicles that becomes the object of exchange in the decentralized, cooperative environment. The results of this scenario’s planning for the current example lead to the following plan.

5.1 Prototypical Implementation

183

Figure 56: Tabular Display of Planning Results for the Decentralized, Cooperative Planning

Figure 56 shows the results of the decentralized, cooperative planning where all three DCs exchange information on their vehicle use and transfer vehicles between them. This scenario results in overall supply chain costs of 2345.29 cost units. Each DC needs two tours to satisfy the demand of four retailers each. DC2 and DC3 require one additional vehicle each; DC3 gets it from DC1 who has two unused vehicles, and DC2 retrieves it from the external LSP. Although DC1 remains with one unused vehicle, the transfer costs (due to the distance) to DC2 are higher (101.97 distance-dependent cost units plus 8 fixed cost units) than the fixed price for a vehicle from the external LSP (75 cost units).

184

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Figure 57: Graphic Display of Results for the Decentralized, Cooperative Planning

As can be seen in Figure 57, this time DC3 retrieves the required vehicle from DC1. Hence, the overall supply chain result is lowered by 16.39 cost units (0.69%). Since the planning is decentralized, the resulting tours are exactly the same as in the non-cooperating environment. Thus, there are retailers that are being delivered by a DC that is located farther away than other DCs. This potential source of cost reductions is addressed by the centralized, cooperative alternative. Here, the SCOptimizer emulates a shift of decision power, and a (virtual) central entity solves a MDVRP. The results are shown in the next figure.

5.1 Prototypical Implementation

185

Figure 58: Tabular Display of Results for the Centralized, Cooperative Planning

Figure 58 shows the results of the centralized, cooperative planning where all three DCs deliver all relevant decision parameters to the (virtual) central entity that makes the decision. In this case, using the Voronoi heuristic, the central entity redefines the business relationships between the DCs and their customers (see Figure 50) and assigns retailers to the DC that are located nearest them. Then, the routing results in eight tours, four of which are performed by DC1, two by DC2, and two by DC3. Since this is a cooperative scenario, a vehicle exchange also takes place, but, since all DCs require all their transportation capacity, the only remaining source of additional vehicles is the external LSP. The overall supply chain results in costs of 1893.78 cost units.

186

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Figure 59: Graphical Display of Results for the Centralized, Cooperative Planning

As can be seen in Figure 59, in this scenario DC1 delivers goods to those six retailers that are located nearer to it. It uses four vehicles (for tours 0, 1, 2, and 3). In addition, the result of the customer reassignment lets DC2 and DC3 supply only three retailers each. They still need two tours, which causes both to retrieve an additional vehicle. Since DC1 now needs four delivery tours, it does not offer any unused vehicles to the other DCs which means that both DC2 and DC3 have to retrieve the needed vehicles from the external LSP. In total, this scenario has better capacity utilization than the other decentralized environments but, on the other hand, it causes higher vehicle costs since it leads to eight (with vehicle costs of 201 cost units) instead of six tours (with vehicle costs of 168.61 in the cooperative and 185 in the non-cooperative decentralized decision) for satisfying all retailer orders. On the other hand, in comparison to the decentralized scenarios, it becomes clear that the total covered distance in the central environment is significantly lower. This fact leads to overall supply chain costs that are 451.51 cost units (19.25%) lower than in the decentralized, cooperative case and 467.91 (19.81%) lower than in the decentralized, non-cooperative case. Table 25 summarizes the results of this exemplary evaluation of the different degrees of cooperative distribution planning.

5.1 Prototypical Implementation

187

Table 25: Summary of Supply Chain Results of the Exemplary Evaluation of Distribution Planning Cooperation Degree

Used Unused Vehicles Vehicles (Total Tours)

LSP Vehicles Required

Fixed Vehicle Costs

Variable Tour Costs

Total Costs

Centralized, cooperative Planning

8

0

2

201

1692.78

1893.78

Decentralized, cooperative Planning

6

1

1

168.61

2176.69

2345.30

Decentralized, noncooperative Planning

6

2

2

185

2176.69

2361.69

Considering the assumptions made and the input parameters and constellation of this example, it can be said that the centralized, cooperative planning leads to better supply chain results. With these figures, planners can aid their decision if and how to implement a cooperation with the distribution partners in the supply chain. For instance, if considering to exchange transportation capacity between the DCs, the potential benefits (16.39 cost units in this example) can be understood as the maximum additional costs that the realization of the information and vehicle exchange might cause. This realization can, for example, mean that the partners have to implement and maintain the corresponding information systems, e.g. an EDI infrastructure. But if the benefits are low, as in this example, the corresponding realization might take the shape of an employee at each DC making calls to the partner DCs and taking the appropriate decision. The evaluation of potential benefits of cooperation with the SCOptimizer might, thus, aid not only the decision if to cooperate, but also how to scale the realization of corresponding support and enabling systems. If the planner considers the realization of centralized cooperation in the supply chain, the potential benefits identified with the SCOptimizer (e.g. 467.91 cost units compared with the lack of cooperation in this example) can also be understood as the upper limit for the expenses in the implementation and maintenance of a planning system that performs the required MDVRP at each planning period and gathers the required information from the collaborating parties. The costs for a corresponding APS for supply-chain-wide distribution planning should, thus, not exceed the identified 467.91 cost units for centralized cooperation to be beneficial to the supply chain as a whole.

188

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Of course, the simplification that is implicit in a model like the one presented here does not allow extrapolating the results to the real context without taking into account further aspects. Examples of such considerations are: •

Transaction costs; mostly in form of coordination efforts between cooperating parties.

•

Implementation and running costs for the information systems to support the cooperation. Here, both the system for running optimization (e.g. an APS) and systems for supporting efficient information exchange (e.g. EDI or XML/EDI systems) are relevant for consideration.

•

Acceptance issues due to cultural or personal issues, for example.

•

Trust issues and further co-opetitive aspects (see Section 2.2.4.1).

•

Motivation of potential partners to cooperate due to either a team understanding (in analogy to the team assumption; see Section 2.3.2), a hierarchical imposure (due to the bargaining position within the supply chain), or an appropriate incentive structure.

This last aspect is of key relevance since often the better global result comprises a worse performance of single partners. In the example presented above, DC1 performs better in the decentralized environments than in the centralized scenario. Table 26 shows the individual performance of distributors in this example. Table 26: Individual Performance of Cooperating Partners Cooperating partner

Centralized, Cooperative Planning

Decentralized, Cooperative Decentralized, NonPlanning Cooperative Planning

DC1

862.49

816.66

816.66

DC2

426.84

724.17

724.17

DC3

604.45

804.47

820.86

Supply Chain

1893.78

2345.30

2361.69

As Table 26 shows, DC2 and DC3 are the “winners” in the centralized, cooperative planning. If the incentive structure is not well designed, there is apparently no rational reason for DC1 to enter the cooperation since it performs 45.83 cost units worse than in the decentralized scenarios. It might be assumed that the supply chain-centric model of competition would encourage such cooperation. The reality, though, as already mentioned, offers a picture in which the firm-centric view of competition dominates the supply chain environment. Game-theoretical aspects must certainly be considered by decision makers in order to enact such a cooperating scenario. If there

5.1 Prototypical Implementation

189

are no dependencies or market position of the partners, then there is a need for a proper incentive structure in order to motivate DC1 to take part in a central cooperation. In addition, the only “winner” in the decentralized, cooperative planning scenario is DC3. In such a constellation, it might be doubtful if DC1 and DC3 want to enter a collaboration at all since there are further aspects (see above) that work against the collaborative environment. In a constellation like the one presented in Table 26, it might be difficult for DC2 to create an incentive structure for the other DCs to participate in the cooperation since the benefits are particularly low (in the current example only 0.69% of the total supply chain costs). What the example described in this section indicates is that a non-cooperative environment will probably lead to higher total costs due to two main reasons: First, DCs supply their customers regardless of their location that might be nearer to another DC – as seen in our example – which leads to higher distance-dependent costs. Second, the possibility of getting vehicles from other DCs if the transfer is less expensive than the external source of transportation capacity does not exist. The example shows also that the decentralized, cooperative environment will probably reduce the fixed vehicle costs due to the exchange of information on unused and additionally needed vehicles between partners. Finally, regarding the centralized cooperation, it can be said that the example shows two tendencies: First, the distance-dependent variable costs will probably be reduced due to the reassignment of retailers to the nearest DC. Second, although the possibility exists to exchange vehicles between DCs (as in the decentralized, cooperative scenario), the centralized scenario might lead to higher fixed vehicle costs. These tendencies are contradictory and negate assumptions regarding the advantages of the different degrees of cooperation. Because of this fact, Section 5.2 presents a calculation study that, on the basis of 270 cases, develops some more generalizable statements in the context of the evaluation of the benefits of cooperative distribution planning with the SCOptimizer. Before that, Section 5.1.3 describes another planning environment implemented within the SCOptimizer prototype that also applies the design patterns described above.

190

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

5.1.3 Prototypical Implementation of the Evaluation of the Bullwhip Effect 5.1.3.1 Planning Background It has been recognized in many diverse markets that demand variability increases as one moves up a supply chain from the customers through the suppliers. This so-called bullwhip effect has serious cost implications. Above all, it leads to increased inventory and material costs as well as increased manufacturing expenses due to excess capacity, inefficient utilization of capacity, and overtime (Lee et al., 1997a, p. 547; Simchi-Levi et al., 2002, pp. 101-104). The main causes of the bullwhip effect are primarily a consequence of the distortion of information on the actual end customer’s demand throughout the supply chain (see Section 2.2.4.2). The forecasting methods and also the reaction of actors to expected supplier shortfalls lead to this distortion. Furthermore, price variations and order batching can also be a cause of variance of demand amounts that are often not foreseeable for the companies upstream in the supply chain (Lee et al., 1997a; Baganha & Cohen, 1998; Cachon, 1999). The effects of these factors appear independently but enhance each other causing the bullwhip effect to increase. After having described the causes of the bullwhip effect in Section 2.2.4.2, the next paragraphs will show how to quantify this effect with the SCOptimizer. For this purpose, the focus is set in a very simplistic supply chain, i.e. a two-stage supply chain consisting only of a retailer who observes end customer demand, and a manufacturer who receives orders from the retailer (Simchi-Levi et al., 2002, p. 104). The discussion is built around the impact of demand forecasting on the bullwhip effect as described by Chen et al. (2000) and Simchi-Levi et al. (2002, pp. 104ff). Applying the different degrees of cooperation presented in Section 2.3.4, the decentralized, noncooperative case is characterized by the independent creation of forecasted demand plans from customers which serve as the basis for the creation of orders of each company in the supply chain. Customer demand is defined as a random figure with an independent distribution at each period. In this model, lead times (L) are assumed to remain constant and all participating companies pursue the same inventory policy (an (s,S) inventory policy). That means, companies only place an order if inventory levels fall below the reorder point s. The amount ordered equals

S minus the residual inventory (Sylver et al., 1998, pp. 238f). Further assumptions concern the forecasting method. All participants perform a simple moving average forecasting with which they calculate both customer demand and order-up-to-point (i.e. the target inventory level) for the next period (for an overview and discussion of additional forecasting methods, primarily exponential smoothing, refer to Chen et al., 2000a; see also Zhao et al., 2002).

5.1 Prototypical Implementation

191

In a first step, the focused supply chain faces a retailer that observes the end customer’s demand Dt of a specific product at period t. and subsequently orders an amount qt of this product from the corresponding manufacturer. As already mentioned, the customer demand is a random figure that satisfies the following formula (Chen et al., 2000, p. 437):

Dt = µ + ρDt −1 + ε t

(5.9)

Here, ȝ is a nonnegative constant and the correlation parameter ρ satisfies |ρ| < 1. This means that the demands are the result of a weak stationary process in which average and variance are constant throughout the time and covariance from two periods is dependent on the distance between these periods and not on the actual time (Schlittgen & Streitberg, 2001, p. 39). Furthermore, the error term εt represents an independent, identical, and symmetric distributed variable with mean 0 and a variance ı2. Thus, the following applies:

E ( Dt ) =

µ 1− ρ

Var ( Dt ) =

σ2 1− ρ2

(5.10)

(5.11)

For ȡ = 0, demand is independent and identically distributed and has an expected value of ȝ and a variance of ı2. The retailer calculates its order-up-to-point yt as follows: yt = Dˆ tL + zσˆ etL

(5.12)

Dˆ tL being an estimate of the expected demand during the lead time and σˆ etL an estimate of the

standard deviation of the error terms during the lead time. The parameter z describes the desired service level (i.e. if safety stocks are to be considered).112

112

Chen et al. (2000, p. 437) point out that the calculation of the reorder point is based on the standard deviation of the forecasting error σ eL whose estimator is σˆ etL and not based on the standard deviation of the demand in the lead time σ L whose estimator is σˆ tL . Although there is a simple relationship between both terms ( σ eL = cσ L , with c being a constant c ≥ 1), it is still appropriate to use σˆ etL (refer to Hax & Candea, 1984, p. 194, for the proof).

192

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

In each period, the expected demand during the lead time is estimated with the simple moving average method with Dˆ tL = Lµˆ t (where µˆ t is the average demand in the last p periods). Thus, the following applies:

§ p ¨ ¦ Dt −i Dˆ tL = L¨ i =1 ¨ p ¨ ©

· ¸ ¸ ¸ ¸ ¹

(5.13)

Furthermore, the estimate of the standard deviation of the forecasting error during the lead time

L is calculated as follows: p

σˆ etL = C L ,ρ

¦ (e i =1

t −i

)2

(5.14)

p

Here, CL,ȡ is a constant function of L, ȡ and p, and et is the one-period forecast error (i.e. et = Dt − Dˆ t1 ; Chen et al., 2000, p. 437). Now, the order amount qt of the retailer can be defined

as:

qt = yt − yt −1 + Dt −1

(5.15)

In the next step, yt and yt-1 can be estimated as shown in (5.12) and inserted in (5.15) (Chen et al. 2000, p. 438):

(

)

qt = Dˆ tL − Dˆ tL−1 + z σˆ etL − σˆ eL,t −1 + Dt −1

(5.16)

This is the formula with which the order amount of each company can be calculated for each period. As Chen et al. (2000, p. 437) point out, qt may be negative in which case it can be assumed that the excess inventory can be returned without additional cost. Based on these assumptions and the corresponding formulas, the bullwhip effect (as the increase in demand variation from one echelon of a supply chain to the next) can be determined by calculating the ratio of the demand variance faced by the preceding member (e.g. demand of the retailer from the manufacturer), Var (q) , and the demand variance faced by the succeeding member (e.g. demand

5.1 Prototypical Implementation

193

of the end customer from the retailer), Var (D) (Simchi-Levi et al., 2002, p. 105). From (5.16) follows:

(

)

qt = Dˆ tL − Dˆ tL−1 + z σˆ etL − σˆ eL,t −1 + Dt −1

§ Dt −1 − Dt − p −1 · ¸¸ + Dt −1 + z σˆ etL − σˆ eL,t −1 = L ¨¨ p ¹ ©

(

)

§ L· L = ¨¨1+ ¸¸ Dt −1 − Dt − p −1 + z (σˆ etL − σˆ eL,t −1 ) p © p¹

(5.17)

The next step leads to the variance of qt: 2

2

§ L· §L· L§ L· Var (qt ) = ¨¨1 + ¸¸ Var (Dt −1 ) − 2 ¨¨1 + ¸¸ × Cov (Dt −1 , Dt − p −1 ) + ¨¨ ¸¸ Var (Dt − p −1 ) p© p¹ p¹ © © p¹

(

)

(

§ L· + z 2 Var σˆ etL − σˆ eL,t −1 + 2 z ¨¨1 + 2 ¸¸ × Cov Dt −1 ,σˆ etL p¹ ©

)

§ 2 L 2 L2 · § 2 L 2 L2 · § L· + 2 ¸¸ × ρ p Var (D ) + 2 z ¨¨1 + 2 ¸¸ = ¨¨1 + + 2 ¸¸ Var (D ) − ¨¨ p p p p p¹ © ¹ ¹ © © × Cov Dt −1 ,σˆ etL + z 2Var σˆ etL − σˆ eL,t −1

(

)

(

(

)

)

(

º · L· § ¸¸ 1 − ρ p » Var (D ) + 2 z ¨1 + 2 ¸ Cov Dt −1 ,σˆ etL P © ¹ ¹ ¼ 2 L L + z Var σˆ et − σˆ e,t −1

ª § 2L 2L = «1 + ¨¨ + 2 p ¬ © p

2

(

)

) (5.18)

with § ρp · 2 ¸ ⋅σ Cov (Dt −1 , Dt − p −1 ) = ¨¨ 2 ¸ ©1− ρ ¹

(5.19)

and

Var (Dt ) =

σ2 1− ρ2

(5.20)

194

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Now, considering the assumptions described above, if the estimate of the standard deviation of the forecasting error σˆ etL (as defined in (5.14)), then, the following applies:113

(

)

Cov Dt −i ,σˆ etL = 0

∀i = 1,..., p

(5.21)

Consequently, the lower bound of the variance of the orders, Var (q) , placed by a member (e.g. the retailer) to the preceding member (e.g. the manufacturer) satisfies § 2 L 2 L2 · Var (q ) ≥ 1 + ¨¨ + 2 ¸¸ ⋅ 1 − ρ p . Var (D ) p ¹ © p

(

)

(5.22)

with lead time L and p as the amount of periods used for calculating the average for the simple moving average forecasting method. The bound is tight for z = 0. Assuming there is no correlation between orders at any given time (i.e. ρ = 0), the following applies:114 Var (q ) 2 L 2 L2 ≥ 1+ + 2 Var ( D) p p

(5.23)

Thus, the variability of the demand is higher, the longer the lead times are and the fewer periods are used for calculating both the forecasted demand and the order-up-to-point. (Simchi-Levi et al., 2002, pp. 105f). For example, if p = 5 and L = 2, then

Var (q) ≥ 2.12 which means that the Var ( D)

variance of orders placed by the retailer to the manufacturer is at least 112 percent larger than the variance of customer demand observed by the retailer. If p = 10 and L = 1, then

Var (q) ≥ 1.22 Var ( D)

which means that the variance of orders placed by the retailer to the manufacturer is at least 22 percent larger than the variance of customer demand observed by the retailer. (5.23) shows that the bullwhip effect is magnified if lead time increases and p decreases (Simchi-Levi et al., 2002, p. 105). If the bullwhip effect is at least partly caused by the distortion of demand information, one could assume that one solution to the problem lies in improving the coordination of information as well as in improving planning along the supply chain (Lee et al., 1997b, see also Section 2.2.4.2).

113

Chen et al. (2000) refer to the proove presented in Ryan (1997) and Chen et al. (1998).

114

Here, there are no safety stocks considered, i.e. z = 0.

5.1 Prototypical Implementation

195

The reason for this is that if information on actual customer demand were available to each stage of the supply chain, all parties could create more accurate forecasts (Simchi-Levi et al., 2002, p. 110). Simchi-Levi et al. (2002, pp. 109f) list also further methods for coping with the bullwhip effect. First of all, variance inherent in the customer demand process should be reduced, e.g. by foregoing price promotions. Another partial cure for the bullwhip effect could be the reduction of lead time whereby lead time usually is the sum of information lead time (i.e. the time needed for processing an order) and order lead time (i.e. the time needed for producing and shipping an item). Hence, lead time could, for instance, be reduced by the use of information technology such as web-based applications or EDI applications. Last but not least, the “bullwhip effect problem” could be controlled by forming vertical strategic partnerships because such cooperative relationships usually involve the sharing of information which might contribute to diminishing the bullwhip effect. The following paragraphs will show how the performance of a multistage supply chain can be improved through cooperation. In this context, the improvement of performance is measured in terms of the reduction of the bullwhip effect through decentral or central cooperation, i.e. through the exchange of demand data between the different stages of the supply chain. Case 1: Supply Chain with k Levels and No Cooperation The described model of a two-stage supply chain is to be generalized and a supply chain with k levels is assumed. Moreover, it is assumed that the succeeding level of the supply chain does not demand information available to the preceding level. Therefore, each party must make estimates of the mean demand based on orders received from the succeeding level of the supply chain. Each party may use a different forecasting technique and employ different buying practices. One possible technique is, once again, the moving average technique with p observations. Each party determines the order-up-to-point based on its demand forecast and accordingly places orders to its supplier (Simchi-Levi et al., 2002, p. 107). In such a supply chain, variance of orders placed at the kth echelon of the supply chain, Var (q k ) , relative to variance of end customer demand,

Var (D) , satisfies

( )

k ª 2 L 2 L2 º Var q k ≥ ∏ «1 + i + 2i » Var (D ) i =1 ¬ p p ¼

∀k

(5.24)

with Li being the lead time between two stages of the supply chain; that is, between stage i and i + 1 (Chen et al., 1998; Simchi-Levi et al., 2002, pp. 107f; for the derivation see Chen et al., 2000,

196

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

p. 439). (5.24) indicates that “variance increases multiplicatively at each stage of the supply chain” (Simchi-Levi et al., 2002, p. 108), in other words, that variance rapidly increases the farther one moves up the supply chain. This is demonstrated by the following example:

( )

If k = 2, p = 5 and Li = 1 for all i, then

Var q 2 ≥ 1.48 Var (D )

If k = 3, p = 5 and Li = 1 for all i, then

Var q 3 ≥ 2.19 Var (D )

( )

(5.24) demonstrates that in a supply chain with k stages and no cooperation, variance increases rapidly the farther one moves up the supply chain: Between the first and second level it grows by 48 percent, and between the first and third level by 119 percent. Case 2: Supply Chain with k Levels and Decentral Cooperation Now a supply chain with k levels and decentral cooperation will be assumed. Decentral cooperation means that the succeeding party always passes along information on (forecasted) mean end customer demand to the next party farther up in the supply chain. Each party may use a different forecasting technique and may employ different buying practices (Simchi-Levi et al., 2002, p. 107). One possible technique is, once again, the moving average technique with p observations. All parties determine their order-up-to-point based on the forecasts of end customer demand and accordingly place orders to their supplier (Simchi-Levi et al., 2002, p. 106). If k = 4, planning and cooperation could look as follows: “[The] retailer, or the first stage in the supply chain, observes customer demand, forecasts the mean demand using a moving average with [p] demand observations, finds his target inventory level based on the forecast mean demand, and places an order to the wholesaler. The wholesaler, or the second stage of the supply chain, receives the order along with the retailer’s mean demand, uses this forecast to determine his target inventory level, and places an order to the distributor. Similarly, the distributor, or the third stage of the supply chain, receives the order along with the retailer’s forecast mean demand, uses this forecast to determine his target inventory level, and places an order to the fourth stage of the supply chain, the factory” (Simchi-Levi et al., 2002, pp. 106f).

In this case, variance of orders placed at the kth stage of the supply chain, Var (q k ) , relative to the variance of end customer demand, Var (D) , is

5.1 Prototypical Implementation

197

k

( ) k

Var q ≥ 1+ Var (D )

2¦ Li i =1

p

§ k · 2¨ ¦ Li ¸ + © i =1 2 ¹ p

2

∀k

(5.25)

with Li being the lead time between two stages of the supply chain, i.e. between stage i and i + 1 , and p being the number of observations in the moving average (Simchi-Levi et al., 2002, p. 107; for the derivation of (5.25) see Chen et al., 2000). The bullwhip effect is smaller than in Case 1 (no cooperation) because the retailer’s forecast of mean end customer demand is made available to all other members of the supply chain. This enables all companies to determine their order-upto-point based on this mean demand which reflects actual end customer demand more precisely than, for example, the retailer’s mean demand forecasted by the wholesaler. (5.25) assumes that there are no safety stocks at any stage of the supply chain. In a multistage supply chain with decentral cooperation “the increase in variability at each stage is an additive function of the lead time and the lead time squared, while for supply chains without centralized information, the lower bound on the increase in variability at each stage is multiplicative” (Chen et al., 2000, p. 442). The following example demonstrates the increase in variance from stage to stage:

( )

If k = 2, p = 5 and Li = 1 for all i, then

Var q 2 ≥ 1.48 Var (D )

If k = 3, p = 5 and Li = 2 for all i, then

Var q 3 ≥ 2.12 Var (D )

( )

(5.25) shows that variance increases the farther one moves up in the supply chain; however, the comparison of variance at k = 3 in a supply chain with decentral cooperation and variance at k = 3 without cooperation shows that this increase is smaller in the scenario with decentral cooperation (119 percent versus 112 percent) because in this case each party is in a better position to accurately forecast mean end customer demand. This observation leads to the following conclusion: “[Centralizing] demand information can significantly reduce, but will not eliminate the bullwhip effect” (Simchi-Levi et al., 2002, p. 109). Case 3: Supply Chain with k Levels and Central Cooperation Next a supply chain with k levels and central cooperation is assumed. Central cooperation means that a central organization plans and places orders for all members of the supply chain based on the same demand data. Moreover, one forecasting technique and one buying practice is employed

198

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

system-wide. In the other two cases, it would probably be a coincidence if all members employed the same forecasting technique and the same buying practice because the members of a supply chain are independent entities and free to choose their techniques and practices. That part of the bullwhip effect which can be contributed to the distortion of end customer demand due to demand forecasting by each supply chain member is avoided (Buxmann et al., 2003, p. 510). This might lead to a five to ten percent increase of profits as Metters (1997, p. 91) found out through simulation. The total bullwhip effect might further be lowered in this case if lead times were reduced through cooperation (Markland et al., 1998) and if price variability were reduced, e.g. by agreeing on steady pricing policies in the supply chain (Buxmann et al., 2003, p. 510). Generally, the cost reductions resulting from information sharing in supply chains (see also Cachon & Fisher, 2000; Lee et al., 2000; Raghunathan, 2001; Thonemann, 2002) represent the value of cooperation which has to be compared with the costs of implementing corresponding support structures and information systems (e.g. SCM software). As a conclusion, one could say that with regard to the bullwhip effect, the constellation “central cooperation” might be the most preferable one because in this case the bullwhip effect is probably the least when compared to the other alternatives (Simchi-Levi et al., 2002, p. 111) and, thus, the performance of the supply chain is probably the highest. The reason for this is that one of the major causes of the bullwhip effect, demand forecasting at each echelon, is eliminated. Even though this constellation seems to be the superior form of cooperation, it might, however, have little relevance for SCM practiced by companies today because it can be assumed that hardly any company would be willing to sacrifice its autonomy and let another party plan for it for the good of the chain as a whole. Furthermore, the complexity of centrally managing an entire supply chain would probably be prohibitive. Implementing “decentral cooperation” seems to be more realistic and, as has been shown in this section, this can facilitate a substantial reduction of the bullwhip effect and, thus, its adverse effects on the supply chain as a whole as well as its members compared to the case with no cooperation. The next section gives an overview from the planner’s perspective on how the evaluation of the benefits of cooperation in the context of the bullwhip effect can be identified through the use of the SCOptimizer.

5.1.3.2 Description from the Planner’s Point of View In analogy to the distribution scenario described in Section 5.1.2, the evaluation of the influence of cooperation on the variation of demand in a supply chain has been exemplarily implemented

5.1 Prototypical Implementation

199

in the SCOptimizer prototype.115 This evaluation is based on the model presented in the preceding section and offers the planner the possibility of comparing the three cooperative strategies presented above: •

Decentralized, non-cooperative scenario. In this case, the actors involved in the planning perform their forecasting using former demand data from only their direct customer. Based on these historic figures, each company anticipates the demand data and creates its own purchase plan.

•

Decentralized, cooperative scenario. In this case, each actor receives information on the end customer’s demand from the point of sale (POS) in the supply chain: the retailer. Based on this data, each actor performs the forecasting considering the lead times between echelons of the supply chain.

•

Centralized, cooperative scenario. This alternative involves a (virtual or existing) central entity that performs the forecasting for all actors and synchronizes the demands at each echelon, considering lead times between each actor. This central entity has historic data on the end customer’s demand and, above that, all additional relevant parameters for the planning.

The first step of the evaluation using the SCOptimizer involves the design of the scenario in which the bullwhip effect and its consequences for the supply chain will be quantified. In analogy to the distribution scenario, the planner is able to use the modeling mask of the SCOptimizer to define the structure of the supply chain as well as the flow of goods and information between the actors. Figure 60 shows the modeling mask of the SCOptimizer containing a linear supply chain model. This supply chain will serve as the basis for the example used in this section to describe the planner’s perspective when evaluating the consequences of the bullwhip effect. It consists of 6 echelons (from “Tier 2 Supplier” through “End Customer”) and considers the possibility of the retailer passing on its historic customer demand figures to the upstream companies (red arrows that point down in Figure 60). This transfer of information is only done in the decentralized, cooperative context in which each actor remains in possession of planning power while exchanging relevant information. The flow of goods is defined by the black arrows that point up in Figure 60.

115

For the complete source code and the files involved in implementing the solving class for the bullwhip effect see Appendix C.4. For all the files involved in the exemplary evaluation of the bullwhip effect see Appendix C.2.

200

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Figure 60: Exemplary Model for the Evaluation of the Bullwhip Effect

After creating the model, the planner proceeds to select “Logistics/Bullwhip effect” in order to begin with the evaluation.

Figure 61: Selecting a Planning Method for the Evaluation of the Bullwhip Effect

Figure 61 shows the mask for selecting an appropriate planning method for the evaluation of the bullwhip effect. The example described in this section contemplates a scenario similar to the Audi supply chain (see Section 4.2.1) in which there is only one product flowing from one

5.1 Prototypical Implementation

201

echelon to the next. It is also assumed that in the manufacturing process of the demanding company there is only one input product unit per output product unit required (in analogy to the Audis supply chain in which one water pump needs one thermostat, one engine needs one water pump, and one car needs one engine). Considering this, the planner will select the option “One path/one product/Simple Moving Average” as shown in Figure 61. As mentioned above, the SCOptimizer creates this mask at runtime, checking the existence of implemented planning classes and parsing and analyzing their corresponding XML description files (see Figure 62).

This method simulates the bullwhip effect in a single path supply chain focusing only on the forecasting error. This is a supply chain with just one element at each level (e.g. one supplier, one manufacturer, one retailer, etc.). Every stage of the supply chain (excepting customers) follows a simple order-up-to inventory policy, i. e. at the end of every period the inventory is accumulated to the order-up-to-point. Furthermore, we assume that all members of the supply chain use the simple moving average to estimate the demand and to schedule the order-up-to-point. In this method, just one product is considered, meaning there is no bill of materials that explodes trough the levels of the supply chain.

true false false false true true

How many periods in the past are considered for forecasting the demand for the simple moving average forecasting method

The demand of the end customer of this supply chain over a certain amount of periods.

…

Figure 62: Excerpt from the Description File for the Bullwhip Effect Evaluation

In analogy to the description of the planning methods for distribution (see Figure 42), this description defines relevant input data that are to be entered by the planner (“InputParam” tags) and additional options to be shown as input masks by the prototype (“optiondlg” tag). The additional options require the planner to enter the bill of materials (BOM) (“bom” tag is set true), cooperation form (“cooperationform” tag), and the number periods for which there exist historic demand data which are to be considered in the forecasting with the simple moving average method (“lengthoftimeseries” tag). After selecting “Next,” the system parses this description file and creates the following mask:

202

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Figure 63: Option Dialog with Required Input from the Planner

By pressing the corresponding button, the planner can access the mask for entering the required information prior to the detailed actor-dependent input. As mentioned above, the selected evaluation method requires the planner to define the cooperation degree which is to be planned.

Figure 64: Input Mask for Selecting the Degree of Cooperation

Figure 64 shows the corresponding masks in which the planner has selected to run a decentralized, cooperative planning. As mentioned above, this option enables actors to pass on their historic demand data to other companies upstream in the supply chain. In contrast to the distribution scenario (see Figure 52), the bottom part of this mask does not require one to select cooperating partners but to determine the actor in the supply chain who is to publish its demand data to other companies upstream in the supply chain. In the example presented here, the model created the possibility for the retailer to pass on its historic demand information to all other companies (see Figure 60 and the information connections in the model); thus, the planner will check the retailer in this mask. Later when the planning is triggered, the SCOptimizer analyzes existing information connections and simulates the exchange of information in the decentralized, cooperative evaluation scenario. In this mask, the planner is able to select any combination of the

5.1 Prototypical Implementation

203

three cooperation degrees (see the upper part of the mask in Figure 64) in order to let the planning class perform a batch run. A further input requirement is shown in the next figure.

Figure 65: Input Mask for Determining the Characteristics of Historic Demand Data

As Figure 65 shows, the model requires the definition of the characteristics of historic data for forecasting. Firstly, the planner must enter the amount of periods that will be the basis for evaluation (16 periods in this example). Secondly, the planner can opt to define the demand observed during these periods either as an isolated period of time with no past values assumed (see the top check box in Figure 65), or as a part of a never ending demand cycle which always has the values declared in these 16 periods. Finally, the bottom check box in Figure 65 lets the planner determine if excess demand of a product may be returned to the supplier without additional cost. This assumption allows the planning model to have negative demand in certain periods that usually follow periods during which the company ordered too much. If the planner leaves this check box unchecked, the lowest possible order amounts to zero product units. If checked, the demand variation is higher than if unchecked. For this example, the demand is assumed to be part of a never ending cycle and allows negative demand (i.e. companies may return excess inventory without any additional cost).

Figure 66: Input Mask for Selecting the Corresponding Bill of Materials

Figure 66 shows the last input mask of the option dialog in which the planner enters the path of a BOM file. This file must be an XML file with a predefined structure which is then parsed by the planning class in order to determine the dependencies between products in the supply chain. Figure 67 shows an imaginary BOM file for an Audi A4 model. There, each product has an id (“id” attribute in the “product” tag) and lists all input products with its corresponding amounts (“contains” tag and its sub-elements “product”). With this document structure, the planner can define the characteristics of BOMs from the end product (“AUDI_A4” in Figure 67) through to the initial input factors with no further decomposition

204

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

(“EAN111”, “EAN222”, and “EAN333” in Figure 67) In analogy to the model of Chen et al. (2000), the example presented here does not require a BOM which means that an imaginary product (“EAN123” in this case) is passed from one echelon to the next with no decomposition of input factors at all.

Audi A4 Car model of the middle class with a sports-oriented focus

1 1 1

Motor for Audi A4 Diesel Motor for Audi A4

1

Cockpit for Audi A4 Sport Cockpit for Audi A4

1

Carrocery for Audi A4 Carrocery for Audi A4

1

Material EAN111 Material for Motors

Material EAN222 Material for Cockpits

Material EAN333 Material for Carroceries



Figure 67: Bill of Materials for a Single-Path Supply Chain

After entering this model-dependent data, the planner can access the actor-dependent input masks by pressing “Next” in the option dialog (see Figure 64). At this stage, the SCOptimizer parses the description file of the planning method (see Figure 62) and applies the definitions of the input parameters’ description:

5.1 Prototypical Implementation

205



nonCustomers

Name

Amount of Periods





customer

timeSeries

SupplyName

Figure 68: Excerpt from the Description File for the Input Parameters for the Bullwhip Effect Evaluation

As can be seen in Figure 68, this file describes the register tabs in the input mask (“Tab” tags and its corresponding attribute “name”). Inside each tab, the file defines which actors require this input data (“listSource” tag) and the data type of the corresponding input fields (“field” tag and its corresponding attribute “type”). The tags “table” define the structure of the input table to appear in the mask. The resulting mask can be seen in the next figure.

Figure 69: Input Mask for Actor-Dependent Parameters for the Bullwhip Effect Evaluation

Figure 69 shows the input mask for the actor-dependent data on product “EAN123”. The first tab lets the planner determine the forecasting horizon of each company involved in the evaluation. This horizon (5 periods in this example, i.e. p = 5 in the model) defines the number of

206

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

observations that serve to calculate the forecast of each period (a result of the average of these observed periods). The second tab (“Demand”) lets the planner enter the end customer’s demand for each period. For this example, the following figures apply: Table 27: End Customer’s Demand per Period Period

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

Demand

3

3

5

5

3

3

5

5

3

3

5

5

3

3

5

5

The next tabs allow the planner to determine if safety stocks are to be considered (meaning to be respected or, if necessary, to be created; see “Safety Stock” in Figure 69) and to declare if the actors possess an initial stock of the demanded product during the initial period (see “Initial Stock” in Figure 69). In this example, neither safety stocks (i.e. z = 0 in the model) nor initial inventory is considered. The tab “Lead Time” lets the planner enter the lead times between involved companies. For this example a lead time of Li = 1 for all i = 1, …, k is assumed. This means that orders placed at the end of a period are supplied at the beginning of the next period which is corresponding to no actual lead time.

Figure 70: Planner’s View of the Input Table for Costs Involved in the Bullwhip Effect Evaluation

The bullwhip effect primarily leads to increased inventory holding costs. It can also influence the price of input factors from suppliers due to unplanned deliveries, short-term capacity enhancements, inefficient capacity utilization, increased personnel costs (e.g. extra working hours), and transportation costs (Lee et al., 1997a.). In the exemplary implementation of the

5.1 Prototypical Implementation

207

bullwhip effect evaluation within the SCOptimizer, the planner is able to provide the costs of inventory holding, shortage costs (which are assumed to cover extra deliveries, unplanned production, and all of the involved costs), and order costs (which are to cover the costs of an order process) for each actor (see Figure 70). These figures are the basis of the quantification of the implications of different degrees of cooperation in the context of the bullwhip effect. In this example, all actors have similar costs: 5 cost units per product unit inventory holding costs, 30 cost units per product unit shortage costs, and 2 cost units per order. After entering this data, the planning can be triggered by pressing the “Create Plan” button. The planning class implemented for this example applies the model of Chen et al. (2000) and, in analogy to the distribution scenario, creates a temporary XML file in which all results are stored. Here, the planner is also given the option to permanently store the results in the desk’s directory (see Appendix C.2 for the complete results file of this example). Following the design patterns for planning classes, the SCOptimizer displays the results of the planning both in a graphical and a tabular fashion. Figure 71 shows the graphic display of calculated orders of each participant in the exemplary supply chain in the decentralized, non-cooperative scenario (orders are placed at the end of each period).116 This mask allows the planner to view and combine several aspects of the results. For example, the bottom-left part of the mask allows the planner to select which amounts are to be displayed within the graph. In addition to the order amounts, also forecasted demands, actual demands, and inventories and safety stocks can be displayed within the graph for each actor or for any combination of them. The bottom-right part of the mask offers a set of buttons that allows further display properties. The first three buttons are toggle buttons that active or deactivate the display of an available cooperation degree. In Figure 71, there is only “Decentralized, non-cooperative planning” activated which is why only the chart for this scenario is shown within the graph (see Figure 77 and Figure 79 for the other scenarios). The button “Show detailed data” opens a new mask with the tabular display of results (see Figure 72). The button “Load XML results file” lets the planner load results from another planning run and display them in the same mask but within a new register tab (a tab which is displayed next to the currently active tab at the top of the window “r_ChenEtAl_all_de…”). Finally, the “Close” button closes the window and returns to the modeling mask of the SCOptimizer.

116

Please note that the end of the first period (t = 1) is highlighted with a grey vertical line in the graph. Please also note that as it is assumed that the demand entered by the planner is part of a neverending cycle, the lines have imaginary past values (for t < 1) that behave as the values shown in the periods observed.

208

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Figure 71: Graphic Display of Results of the Bullwhip Effect Evaluation for the Decentralized, Non-Cooperative Scenario117

As can be seen in Figure 71, the initial demand of the end customer (red line with the square shapes) varies between 3 and 5 units. Moving up in the supply chain, the graph shows how the variance of orders increases from a range between 2.6 and 5.4 units for the retailer (blue line with circled shapes) to a range between 0.23 and 7.77 for the orders of the tier 2 supplier (pink line with down looking triangled shapes). This increase is described by (5.22). As already mentioned, in order to view the tabular display of results, the planner has to press the “Show detailed data” button. Figure 72 shows the tabular display of the orders of the decentralized, non-cooperative scenario. This mask also offers a range of display possibilities to allow corresponding analyses. The main navigation is realized by the three dropdown lists at the top of the window. The left list allows one to display values of single actors or of all of them. The center list allows one to select the display values of either single products or of all of them depending on the BOM entered before the planning took place (see Figure 66).

117

The resulting planning figures are discrete values represented by the shapes in the graph. The lines in the graph are just for display purposes.

5.1 Prototypical Implementation

209

Figure 72: Tabular Display of Results of the Bullwhip Effect Evaluation for the Decentralized, Non-Cooperative Scenario

Finally, the right list allows one to select the display of several aggregations of the resulting figures: •

Total Amounts. This option displays the summed amounts of inventory, forecasted demand, actual demand, and orders of actors for all periods planned.

•

Amounts per period. This option displays the amounts of inventory, forecasted demand, actual demand, and orders of actors depicted for each period. This is currently selected in Figure 72.

•

Total Costs. This option displays the total inventory holding costs, total shortage costs, and total order costs of actors, summed up for all periods planned.

•

Costs per period. This option shows the costs of inventory holding, shortage, and order of actors depicted for each period.

By combining these three dropdown lists, the planner is offered a wide range of visualization possibilities for the resulting figures. The resulting tables are shown in the central region of the mask. This central region offers three register tabs with the three cooperation degrees of the SCOptimizer: NON-COOP (currently selected in Figure 72; stands for decentralized, noncooperative planning), DEC-COOP (stands for decentralized, cooperative planning), and CENCOOP (stands for centralized, cooperative planning). Within these register tabs, the planner is able to view the corresponding figures. As Figure 72 shows, the planner has chosen to display the amounts per period of product “EAN123” for all actors. Since DEC-COOP is currently active, the system displays the figures within the decentralized, non-cooperative register tab and populates it with 4 subordinate tabs: “Inventory,” “Forecasted Demand,” “Actual Demand,” and “Orders” (to see the values for the other cooperation degrees, the planner has to first select the corresponding tab (either DEC-COOP or CEN-COOP) and then make the selection in the

210

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

dropdown lists). By selecting the corresponding tab, the planner is able to see all the figures resulting from the current planning. In Figure 72, the planner has activated the “Orders” tab and is able to see the table with the orders of each actor at each period. It can be seen how the variance of orders increases with each echelon one moves up in the supply chain. The bottom part of the mask (see Figure 72) shows a summary of the total results of the supply chain. The left part displays the key parameters of the planning (number of echelons, number of actors, number of products, etc.) and the right part shows a comparison of the key results of all available degrees of cooperation (see also Table 34). If the planner selects to display total amounts, the mask populates the active cooperation tab as shown in the next figure.

Figure 73: Tabular View of Total Order Amounts in the Decentralized, Non-Cooperative Scenario

In the central region of the mask shown in Figure 73, the planner gets the total amounts of inventories, forecasted demands, actual demands and orders for each actor. For each of these categories, the planning class shows them maximum, minimum, and average values as well as the variance, standard deviation, and the total amount of all periods considered. Table 28 summarizes the resulting amounts of the decentralized, non-cooperative scenario. This data depicts the dimension of the bullwhip effect: While the end customer’s demand only varies between 3 and 5 units, the variations in the preceding echelons are clearly higher. In this example, while the variance of the end customer’s demand is Var(x) = 1, the retailer’s is Var(x) =

1.48 and the wholesaler’s Var(x) = 2.19. This goes on through to the tier 2 supplier who is facing a variance of Var(x) = 7.10.

5.1 Prototypical Implementation

211

Table 28: Order Amounts of Each Actor for Each Period in the Decentralized, Non-Cooperative Scenario Period/ Actor

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

x Var

16

End 3 Customer

3

5

5

3

3

5

5

3

3

5

5

3

3

5

5

4

1

Retailer

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

4

1.48

Wholesaler

5.88 4.92 2.12 3.08 5.88 4.92 2.12 3.08 5.88 4.92 2.12 3.08 5.88 4.92 2.12 3.08 4

2.19

OEM

3.27 6.44 4.73 1.56 3.27 6.44 4.73 1.56 3.27 6.44 4.73 1.56 3.27 6.44 4.73 1.56 4

3.24

Tier 1 Supplier

0.93 3.61 7.07 4.39 0.93 3.61 7.07 4.39 0.93 3.61 7.07 4.39 0.93 3.61 7.07 4.39 4

4.80

Tier 2 Supplier

3.85 0.23 4.15 7.77 3.85 0.23 4.15 7.77 3.85 0.23 4.15 7.77 3.85 0.23 4.15 7.77 4

7.10

As to costs, by selecting the corresponding visualization in the dropdown lists the planner is able to see the resulting costs of the planning.

Figure 74: Tabular View of Inventory Costs per Period in the Decentralized, Non-Cooperative Scenario

As can be seen in Figure 74, the planner is able to see the costs per period for the categories “Inventory,” “Shortage,” and “Orders.” Since the last period in which orders are placed is t = 16, the resulting inventory in period t = 17 still has to be taken into account which is why this table includes 17 columns. In analogy to the display of amounts, the selection of total costs in the dropdown lists populates the central region with further key values.

212

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Figure 75: Tabular View of Total Inventory Costs in the Decentralized, Non-Cooperative Scenario

This additional visualization offers the planner the possibility to view the maximum, minimum, and average values as well as the variance, standard deviation, and total costs of each actor for the categories “Inventory,” “Shortage,” and “Orders.” The inventory cost reveals that the wholesaler is facing higher inventories than the other actors in the supply chain. The inventory level of an actor k for each period, I tk , is calculated as

I tk = I tk−1 − Dtk−−11 + Dtk− L , with Dtk−−11 as the actual demand of the direct customers of actor k that has to be shipped in this period and Dtk− L as the order of this actor placed to its suppliers which due to lead time L arrives at the beginning of this period. When taking a look at the inventory levels throughout the supply chain, the picture uncovers some inefficiencies. As Figure 76 shows, most of the time the wholesaler has too much stock of the considered product (blue line with circled shapes) which results in a higher average inventory and higher inventory holding costs. On the other hand, the tier 1 supplier (yellow line with diamond shapes) faces shortages throughout most of the periods which results in higher shortage costs but also lower inventory holding costs. The resulting fluctuating inventory levels, which are often shortages, give a good example of what the consequences of the bullwhip effect are. As described in Section 2.2.4.2, the uncertainty of how customer demand is going to behave makes it often unpredictable for a company to determine what the right inventory level will be in order to be able to face future demand. Table 29 gives an overview of the results of the decentralized, noncooperative scenario.

5.1 Prototypical Implementation

213

Figure 76: Graphic Display of Inventory Levels in the Decentralized, Non-Cooperative Scenario118 Table 29: Results of the Decentralized, Non-Cooperative Scenario of the Exemplary Bullwhip Effect Evaluation Actors

Inventory Holding Costs

Shortage Costs

Order Costs

Total Costs

End Customer

0

0

32

32

Retailer

72

48

32

152

Wholesaler

128

0

32

160

OEM

30.40

443.52

32

505.92

Tier 1 Supplier

0

1241.09

32

1273.09

Tier 2 Supplier

58.43

460.57

32

551

Supply chain

288.83

2193.18

192

2674.01

In the decentralized, cooperative scenario, the retailer publishes the observed customer demand to the other companies in the supply chain. This means that all actors are able to calculate the estimates of the customer demand ( Dˆ tL and Dˆ tL−1 ) based on the same historic data (see (5.13)).

118

The resulting planning figures are discrete values represented by the shapes in the graph. The lines in the graph are just for display purposes.

214

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

The wholesaler, for example, now has two possibilities for determining its order amounts (see (5.16)): It can either use the retailer’s historic data or the end customer’s data to base the forecasts on. Since the retailer’s demand represents a distortion of the end customer’s demand, the wholesaler calculates its forecast based on the latter. In the model of Chen et al. (2000), the wholesaler is able to ignore past orders of the retailer because they are the result of the same forecasting method and the same inventory policy as the wholesaler’s. Thus, the wholesaler knows that the retailer is applying the simple moving average forecasting method onto the historic end customer’s demand data. This means that, if the wholesaler is only considering information on the end customer’s demand, the wholesaler knows that this information is the only decisive parameter for the shape of the actual retailer’s demand. Of course, in the forecast the wholesaler has to take into account that the lead time to the end customer is higher than to the retailer. Thus, companies receiving information on end customer’s demand have to perform the forecast considering the entire lead time as a result of

¦

k −1

i =1

Li . These varying lead times for

estimating the expected customer demand and its variance again causes a bullwhip effect which is, though, lower than in the non-cooperative scenario (see (5.24) and (5.25)). Figure 77 shows the variance of the orders placed by each actor of the supply chain in the decentralized, cooperative scenario. In this case, the retailer’s demand varies exactly the same as in the non-cooperative scenario (blue line with circled shapes). For the other actors, the scenario offers a better picture. For example, the wholesaler’s demand (green line with the upper-looking triangled shapes) ranges between 5.40 and 2.60 units (5.88 and 2.12 in the non-cooperative scenario) and the tier 2 supplier (pink line with the diamond shapes) ranges between 7.00 and 1.00 units (compared to 7.77 and 0.23 before). Table 30 summarizes the resulting amounts of the decentralized, cooperative scenario.

5.1 Prototypical Implementation

215

Figure 77: Graphic Display of Results of the Bullwhip Effect Evaluation for the Decentralized, Non-Cooperative Scenario119 Table 30: Order Amounts of Each Actor for Each Period in the Decentralized, Cooperative Scenario Period/ Actor

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

x Var

16

End 3 Customer

3

5

5

3

3

5

5

3

3

5

5

3

3

5

5

4

1

Retailer

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

4

1.48

Wholesaler

5

2.20 3

5.80 5

2.20 3

5.80 5

2.20 3

5.80 5

2.20 3

5.80 4

2.12

OEM

5

1.8

3

6.2

5

1.8

3

6.2

5

1.8

3

6.2

5

1.8

3

6.2

4

2.92

Tier 1 Supplier

5

1.4

3

6.6

5

1.4

3

6.6

5

1.4

3

6.6

5

1.4

3

6.6

4

3.88

Tier 2 Supplier

5

1

3

7

5

1

3

7

5

1

3

7

5

1

3

7

4

5

Table 30 shows that the increase in variance is lower than in the non-cooperative environment. This is explained by the additive increase of the ratio between end customer’s demand variance

119

The resulting planning figures are discrete values represented by the shapes in the graph. The lines in the graph are just for display purposes.

216

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

and variance of the demand of the corresponding echelon (see (5.25)) compared to the multiplicative increase of the non-cooperative scenario (see (5.24)). The lower increase of variance is reflected in the resulting inventory levels in the decentralized, cooperative scenario.

Figure 78: Graphic Display of Inventory Levels in the Decentralized, Cooperative Scenario120

As Figure 78 shows, in this scenario the retailer faces exactly the same inventory levels as in the non-cooperative scenario (red line with square shapes). For all the other actors, the situation shows a very different picture than before (all other lines and shapes are superimposed and appear to be a single line that varies between 0 and -0.4). Through the direct access to the end customer’s demand, now all actors (from the wholesaler through the tier 2 supplier ) face same inventory levels with an average of -0.18 (i.e. they face a slight shortage on average and a maximum shortage of 0.40). No one has to deal with inventories since the inventory levels are never above zero. The OEM, for example, had to deal with shortages of up to 2.61 units in the non-cooperative scenario and inventories of up to 1.52 units (now it does not have inventories at all). Something similar happens to the tier 1 supplier who increases the average inventory level from -2.30 to -0.18 units per period (i.e. it reduces the average shortage by more than 2 units). In

120

The resulting planning figures are discrete values represented by the shapes in the graph. The lines in the graph are just for display purposes.

5.1 Prototypical Implementation

217

case of the tier 2 supplier, on average, it faces a slightly better situation in that it raised the average inventory from -0.20 to -0.18 (i.e. it lowered the average shortage by 0.02 units). Table 31 gives an overview of the results of the decentralized, cooperative scenario in terms of costs. Table 31: Results of the Decentralized, Cooperative Scenario of the Exemplary Bullwhip Effect Evaluation Actors

Inventory Holding Costs

Shortage Costs

Order Costs

Total Costs

End Customer

0

0

32

32

Retailer

72

48

32

152

Wholesaler

0

96

32

128

OEM

0

96

32

128

Tier 1 Supplier

0

96

32

128

Tier 2 Supplier

0

96

32

128

Supply chain

72

432

192

696

Comparing the overall supply chain result of 696 cost units with the non-cooperative scenario, it can be seen that it is 1978.01 cost units (73.97%) lower. Individually, while the end customer and the retailer stay the same in terms of costs, all other actors are able to lower their total costs. Clearly, the winner in this scenario is the tier 1 supplier who is able to reduce its costs by 1145.09 units which represent a reduction of 89.95%. In the centralized, cooperative scenario the situation is characterized by a central entity that sees all the events and has the exclusive decision making power. As already mentioned, the central entity will try to synchronize orders throughout the supply chain in order to reduce inventory and shortage costs. The results of the planning can be viewed in Figure 79. Figure 79 shows the variance of the orders placed by each actor of the supply chain in the centralized, cooperative scenario. Since the central entity synchronizes the orders of actors (from the retailer’s through to the tier 2 supplier’s demand), Figure 79 appears to only show two lines. In fact, the red line with square shapes shows the shape of the end customer’s demand throughout the relevant periods. The other line is the superimposition of the orders of all other actors which have exactly the same shape. This is the result of the synchronization of the central entity that has taken the results of the retailer’s forecasting and the calculated orders to determine the orders of all other actors under consideration of the lead times. Since in this example, the lead times are all L = 1 between echelons (i.e. orders placed at the end of a period arrive at the beginning of the next period which corresponds to no actual lead time at all), all demand plans

218

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

are the same. Table 32 summarizes the resulting amounts of the decentralized, cooperative scenario.

Figure 79: Graphic Display of Results of the Bullwhip Effect Evaluation for the Centralized, Non-Cooperative Scenario121 Table 32: Order Amounts of Each Actor for Each Period in the Centralized, Cooperative Scenario Period/ Actor

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

x Var

16

End 3 Customer

3

5

5

3

3

5

5

3

3

5

5

3

3

5

5

4

1

Retailer

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

4

1.48

Wholesaler

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

4

1.48

OEM

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

4

1.48

Tier 1 Supplier

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

4

1.48

Tier 2 Supplier

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

5

2.6

3

5.4

4

1.48

121

The resulting planning figures are discrete values represented by the shapes in the graph. The lines in the graph are just for display purposes.

5.1 Prototypical Implementation

219

Table 32 shows that the increase in variance takes only place between the end customer and the retailer since it is assumed the central entity will not take over the decision making power of the end customer. Thus, the central entity faces in a first step the same planning situation as the retailer in the scenarios before: it has to forecast the end customer’s demand. The result of this forecast creates an increase of 48% variance. The resulting demand for the retailer is now synchronized with the demand of all other actors through to the tier 2 supplier which stops a further increase of demand variance. All actors face now a variance of demand of 1.48 units. The resulting inventory levels can be seen in the next figure.

Figure 80: Graphic Display of Inventory Levels in the Centralized, Cooperative Scenario122

As Figure 80 shows, except for the retailer (red line with square shapes), all have no inventories or shortages at all (all other lines are superimposed over the horizontal axis and appear to be a single line). The centralization has allowed all actors to fully synchronize the product flow and thus realize a JIT delivery throughout the supply chain. Table 33 gives an overview of the results of the centralized, cooperative scenario in terms of costs.

122

The resulting planning figures are discrete values represented by the shapes in the graph. The lines in the graph are just for display purposes.

220

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Table 33: Results of the Centralized, Cooperative Scenario of the Exemplary Bullwhip Effect Evaluation Actors

Inventory Holding Costs

Shortage Costs

Order Costs

Total Costs

End Customer

0

0

32

32

Retailer

72

48

32

152

Wholesaler

0

0

32

32

OEM

0

0

32

32

Tier 1 Supplier

0

0

32

32

Tier 2 Supplier

0

0

32

32

Supply chain

72

48

192

312

Comparing the overall supply chain result of 312 cost units with the non-cooperative scenario, it can be seen that it is 2362.01 cost units (88.33%) lower. Individually, while the end customer and the retailer remain the same in terms of costs, all other actors are able to lower their total costs. Again, the actor that comparably performs best is the tier 1 supplier who is able to reduce its costs by 1241.09 units which represent a reduction of 97.49%. In comparison with the decentralized, cooperative scenario, this scenario reduces the overall costs by 384 cost units (55.17%). Individually, again, end customer and retailer remain the same in terms of costs; all other actors are able to lower their total costs by 96 units which represents a reduction of 75% each. Table 34 shows the results of the exemplary evaluation of the bullwhip effect with the SCOptimizer where average figures represent the amounts per period of all actors and total figures represent the sum of all period figures of all actors. Since in this example all actors order during each period in all scenarios, the order costs are the same. It is inherent in this model that the total orders are the same in all three scenarios. Interestingly enough, the centralized and decentralized cooperative scenarios face the same inventory costs. For this model, this means that the lower the shortage costs are in a supply chain, the more similar the overall results of the centralized and decentralized cooperative scenarios are. It has to be noted that the decentralized, cooperative scenario does not underlie the team assumption which puts a question mark to the motivation of the retailer to take part in the implementation of this scenario (considering all the consequences in terms of additional transaction costs and enhanced complexity). Thus, in a real-life implementation of this scenario, it can be assumed that since, for example, the retailer has no direct monetary interest in passing on its customer’s demand information, other actors (primarily the tier 1 supplier) will be

5.1 Prototypical Implementation

221

interested in creating the appropriate incentive structure in order to gain the retailer’s motivation. On the other hand, it can also be assumed that the retailer is interested in a better retail price of the good which is likely to be influenced by the logistics costs such as inventory holding and shortage costs. If lower costs in the supply chain imply a lower price, and this one implies higher margins and revenues, the retailer would have an interest in realizing this scenario. Table 34: Results of the Exemplary Bullwhip Effect Evaluation Key figures

NON-COOP

DEC-COOP

CEN-COOP

Total inventory

57.77

14.40

14.40

Avg. inventory

3.40

0.85

0.85

Total shortage

73.11

14.40

1.60

Avg. shortage

4.30

0.85

0.09

Total order amount

384

384

384

Avg. order amount

24

24

24

Total inventory holding costs

288.83

72

72

Avg. inventory holding costs

16.99

4.24

4.24

Total shortage costs

2193.18

432

48

Avg. shortage costs

129.01

25.41

2.82

Total order costs

192

192

192

Avg. order costs

12

12

12

Total supply chain costs

2674.01

696

312

Avg. supply chain costs

158

41.65

19.06

In contrast to the decentralized, cooperative scenario, the centralized cooperation is characterized by the team assumption which ensures the participation of the retailer in the process. In a real-life implementation, the same thoughts apply as with the decentralized, cooperative scenario with the additional “hurdle” of giving up planning power to an external entity that is to create the demand plan for all actors in the supply chain. The results of the evaluation with the SCOptimizer, though, are to give decision makers additional information on the benefits of different degrees of cooperation which can then be considered for determining which alliances to enter, the dimension of investments in collaboration systems, and the depth of organizational restructuring.

222

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

After describing the architecture of the SCOptimizer and showing the implemented planning classes, the next section shows a computational study of cooperative distribution planning performed with the SCOptimizer.

5.2 Computational Study on Cooperative Distribution – An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer Transportation-related costs can have a great impact on the total supply chain costs (Bergmann & Rawlings, 1998, p. 369; Quinn, 2000, p. 75). For example, Simchi-Levi points out that in “1998, American companies spent about $900 billion on logistics-related activities. Out of this, about 60 percent were transportation related […]. A five percent reduction in transportation costs might have a huge impact on the supply chain. And, of course, this kind of reduction in supply chain costs can have a significant impact on the companies performance”. (Quinn, 2000, p. 75). When evaluating the benefits of cooperation, as presented in this dissertation, it has to be realized that most planning scenarios in management science are often characterized by the stochastic behavior of (several) relevant input parameters. This is the case in the distribution scenario shown in Section 5.1.2.2 since, for example, the actual retailer’s demand is influenced by the end customer’s demand which in reality is not a deterministic measure; it is more likely to be influenced by many aspects that cannot be analytically determined (such as seasonal events, sociological aspects, activities of complementors and competitors, etc.). Furthermore, if, for example, traffic jams and similar transport-dependent events are considered, the variable costs and distances might be subject to uncertainty. As shown in Section 5.1.2.2, the results of a planning round for the considered distribution example do not allow one to formulate general statements on the benefits of the different cooperation degrees. Considering the nondeterministic behavior of demand, sometimes it will be possible to exchange vehicles and sometimes when distributors use all their vehicles due to a higher demand, a vehicle exchange will not take place. The tendencies described within the distribution scenario will sometimes be reinforced and sometimes weaked by the occurring demand. In order to give an insight to the behavior of the results of the different cooperation degrees due to the behavior of the input parameters, this section presents a numerical study that on the basis of 270 cases develops some more generalizable statements. For this purpose, selected planning parameters are combined and random retailer demand is generated. For the thus created cases, the SCOptimizer performs planning runs for each cooperation degree, and the results are

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

223

analyzed and presented in this section. The complete results of the planning rounds can be seen in Appendix E.

5.2.1 Approach of the Computational Study The underlying model of this study (see Section 5.1.2.1) can be defined as a static, deterministic, discrete event model in which a set of input parameters influence the performance of the system at a given point of time (Law & Kelton, 1991, pp. 6ff). The change of state variable with respect to time is not considered (for a definition of the characteristics of the simulation model, see, for example, Law & Kelton, 1991, pp. 6ff and pp. 109ff; Pidd, 1992, pp. 18ff). Although some of the input parameters (namely the retailer’s demand) are probabilistic, there are no stochastic components within the model (e.g. the duration of a route due to possible traffic jams) that would cause it to be non-deterministic. Here, the simulation presents a sensitivity analysis in the form of a combination of different states of input parameters that represent the sample space for the generation of experiments. As mentioned above, this study considers a total of 270 planning cases (i.e. experiments) that provide the sample points of a total of 27 sample spaces (Law & Kelton, pp. 268ff). Thus, for each sample space the author has generated 10 sample cases whose mean values provide the basis for the numbers presented in the analysis. For the definition of the steps of the study, the author adapts the approach of Law and Kelton (1991, pp. 106ff; see also Banks & Carson, 1984, p. 12; Law & McComas, 1990).

Formulation of the problem

Definition of a model and implementation

Documentation and presentation of results

Testing with pilot runs

Valid?

Analysis of output data

Design of experiments

Make production runs

Figure 81: Steps of the Computational Study (Source: Based on Law & Kelton, 1991, p. 107)

As Figure 81 shows, the procedure of the study presented in this section is based upon the formulation of the problem and the definition of the model. The main goal of the study is to depict the consequences of varying state variables of a distribution scenario. The objective is to determine how the performance of the different cooperation degrees is affected by different settings of the scenario described in Section 5.1.2.2. Both the model and implementation are described in Section 5.1.2.2. In analogy to that exemplary evaluation, this study uses the Savings method and a 3-opt improvement for solving the VRP. Although models like this one are a simplification of real situations, they can still provide insight into the tendencies and performance

224

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

of the actual system. In this regard, Law and Kelton (1991, p. 107) point out that a “model should contain only enough detail to capture the essence of the system for the purposes for which the model is intended; it is not necessary to have a one-to-one correspondence between elements of the model and elements of the system”. The following tables show the key parameters of the scenario used for the experiments. Table 35: Key Scenario Parameters for the Creation of Experiments Key Parameters

Values

Distribution Centers

3

Tour Characteristics

Closed

Maximum Tour Costs

Unlimited

Cost Units per Distance Unit

1

Vehicle Capacity

500

External Vehicle Costs

39, 79, 118 (depending on the sample space)123

DC1 Vehicle Costs

8

DC1 Available Trucks

3, 1, 1(depending on the sample space)124

DC2 Vehicle Costs

12

DC2 Available Trucks

5, 3, 3 (depending on the sample space)124

DC3 Vehicle Costs

10

DC3 Available Trucks

4, 2, 2 (depending on the sample space)124

Customers per DC

4

Handling Capacity of DCs

Unlimited

Table 36: Vehicle Transfer Costs between DCs (Fixed plus Variable Costs) From/To DC1

DC2

DC3

DC1

-

109.966

58.606

DC2

113.966

-

65.16

DC3

60.606

63.16

-

123

The LSP vehicle price for the different settings on the Z axis (see Figure 82) is calculated based on the average costs (78.58) from Table 36 and rounded to integer values.

124

The transportation capacity in the supply chain varies depending on the current setting on the X axis as shown in Figure 82.

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

225

The next step (“Testing with pilot runs” in Figure 81 and their validation) can be seen as an example in Section 5.1.2.2. In the opinion of the author, the correctness of the behavior of the planning results validates the model. For the design of experiments, the author faced the challenge of choosing the factors to change in order to depict their consequences in terms of performance of the different cooperation degrees. Considering the number of combinations of settings, the author has selected to simulate three different values of three input factors which combined result in 27 different settings: 3rd Block

Each cell of a block comprises of 10 cases. The values contented in each Cell are average values.

Ce 0.5 1.5

Ce 1.0 1.5

Ce 1.5 1.5

DeCo 0.5 1.5

DeCo 1.0 1.5

DeCo 1.5 1.5

De 0.5 1.5

De 1.0 1.5

De 1.5 1.5

2nd Block Ce 0.5 1.0

Ce 1.0 1.0

Ce 1.5 1.0

DeCo 0.5 1.0

DeCo 1.0 1.0

DeCo 1.5 1.0

De 0.5 1.0

De 1.0 1.0

De 1.5 1.0

1st Block Ce 0.5 0.5

Ce 1.0 0.5

Ce 1.5 0.5

Decentralized Planning with Communication

DeCo 0.5 0.5

DeCo 1.0 0.5

DeCo 1.5 0.5

Decentralized Planning without Communication

De 0.5 0.5

De 1.0 0.5

De 1.5 0.5

Centralized Planning

X

0.5

1.0

Z 1.5

1.0 External Truck Price / Avg. Internal Truck Lead Costs

Y

0.5

1.5

Total Customer Demand / Total Transportation Capacity

Figure 82: Overview of the Experiment Design of the Computational Study

The three factors are: •

Cooperation degree. This factor is shown on the Y axis in Figure 82. The three values are the three cooperation degrees already described above: Centralized, cooperative planning (all sample spaces in this category are characterized by “Ce”), decentralized, cooperative planning (all sample spaces in this category are characterized by “DeCo”), and

226

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

decentralized, non-cooperative planning (all sample spaces in this category are characterized by “De”). •

Demand/capacity ratio. The three values of this factor are shown on the X axis in Figure 82. This factor describes the ratio between the total retailer’s demand (i.e. the sum of demands of all 12 retailers) and the total initial transportation capacity of distributors (i.e. the sum of the capacities of all vehicles at all three DCs). In this case, the author has selected three ratios to evaluate: 0.5 (all sample spaces in this category are characterized by “0.5” after the denomination of the cooperative category), 1.0 (all sample spaces in this category are characterized by “1.0” after the denomination of the cooperative category), and 1.5 (all sample spaces in this category are characterized by “1.5” after the denomination of the cooperative category). This factor seems to be interesting to evaluate in the context of the study since it might influence the number of vehicles needed for satisfying the retailer’s demand and, thus, the availability of internal vehicles to exchange between DCs.

•

External/internal vehicle price ratio. This factor is shown on the Z axis in Figure 82. It describes the ratio between the external vehicle price offered by the LDL and the average costs of transferring one vehicle from one DC to another. Both fixed and variable, transport-dependent costs are taken into consideration. For the experiments presented here, the average is calculated based on the 6 possible transfer relationships between the three DCs. The three values considered for the creation of experiments are: 0.5 (all sample spaces in this category are characterized by “0.5” after the denomination of the demand/capacity ratio category), 1.0 (all sample spaces in this category are characterized by “1.0” after the denomination of the demand/capacity ratio category), and 1.5 (all sample spaces in this category are characterized by “1.5” after the denomination of the demand/capacity ratio category). In the opinion of the author, this factor is worth evaluating since it directly influences the propensity of DCs to reject or accept vehicle transfers from other DCs.

The combination of these factors results in 27 sample spaces for which 10 random sample points (in terms of retailer’s demand quantities) were created. For the generation of the random values of the demand data, it is assumed that the retailer’s demand is a discrete (integer) random variable with a normal distributed probability density function with a mean value of ȝ = 250 and a standard deviation of ı = 100. The author performs a rejection sampling (Pidd, 1992, pp. 206ff; Morgan, 1984, pp. 98ff) in order to achieve the desired ratio between demand and transportation capacity of the distributors. This method, attributed to von Neumann (1951), ensures that all cases fit in the sample space of the corresponding setting within the study. In addition, in order

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

227

to be able to isolate the effects of the different settings in the output performance, the method of common random numbers was used (Law & Kelton, 1991, pp. 613ff). The basic idea of this variance-reduction technique is to be able to “compare the alternative configurations ‘under similar experimental conditions’ so that we can be more confident that any observed differences in performance are due to differences in the system configurations rather than to fluctuations of the ‘experimental conditions’” (Law & Kelton, 1991, pp. 613-614). This means that, for example, in all settings with a demand/capacity ratio of 0.5, the same 10 randomly generated demand settings were used. Because this method relies on an induced positive correlation for its effect, the later analysis of results does not, of course, consider those correlations between sample cases (Morgan, 1984, p. 172). Besides the factors shown in Figure 82, there are further ones that could have served as categorization basis for the experiments, such as the ratio between the average distance of customers to their own distributor and the average distance to the nearest other distributor. Also, distances could have also been a probabilistic variable which would have described the stochastic aspects of a shipment which is often subject of interruptions or delays (e.g. accidents, traffic jams, detours, etc.). Further settings could have been based on the amount of customers per DC, the ratio between fixed and variable costs, variable costs as a random variable (describing, for example, that the consumed fuel is often dependent on factors such as the driving abilities of the corresponding driver, on the weather, etc.). But, as Law and Kelton (1991, p. 109) point out, often “there are more alternatives than one can reasonably simulate” and the author chose the three factors shown above to base the creation of experiments on. The next step towards the completion of the study is the creation of production runs. After the generation of 270 experiments, the author produced the corresponding input data as XML files (see Section 5.1.3.2, especially Figure 47) and developed batch files that controlled the running of the planning for the different categories (for the settings of all experiments see Appendix E.2 (centralized, cooperative planning), Appendix E.3 (decentralized, cooperative planning), and Appendix E.4 (decentralized, non-cooperative planning)). The experiments were performed with the SCOptimizer prototype in which the graphic user interface was deactivated (for the files involved in the planning scenario within the SCOptimizer see Appendix E.1). The scenario of the planning had the same structure as the one presented in Section 5.1.3.2 in which three DCs supply four retailers each. The distances are the same and vehicle fleet and demand varies depending on the sample space of the experiment (see Appendix E.2 through E.4). As already described in Section 5.1.3.2, the MDVRP was solved using the Voronoi heuristics followed by

228

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

the Savings method with a 3-opt improvement algorithm. The VRP was solved using the Savings method and a 3-opt improvement (see Section 5.1.3). Finally, the analysis of the data and its documentation is shown in the next section and in Appendix E. All results of the experiments can be seen in Appendix E.2 through E.4. In addition, the analysis of the resulting figures was performed within a spreadsheet. Since the complete analysis is contained within a total of 9,111 tables, the next section presents only a set of selected results which in the opinion of the author show the most interesting aspects. For the complete analysis see Appendix E.5.

5.2.2 Selected Results of the Computational Study The intention of the results presented in this section is to give an impression of how the different settings, pertaining to both demand/capacity ratio and external/internal vehicle costs ratio, affect the quality of the cooperation strategy primarily in terms of costs. For this purpose, the analysis involved a number of measures that are listed below: •

Total costs,

•

total vehicle costs,

•

total distance (i.e. total variable costs since one distance unit causes one cost unit),

•

total load (only of interest in the individual analysis since the total shipped load is always the same),

•

total costs of vehicle transfer between DCs,

•

total external vehicle costs (i.e. expenses in vehicles from LSP),

•

total tours (i.e. total vehicles needed),

•

external vehicles needed (from an LSP),

•

unused internal vehicles (from the initial fleet of each DC),

•

unused capacity of vehicles used,

•

total unused capacity (including the capacity of unused internal vehicles),

•

total number of vehicles transferred (including both internal vehicles and vehicles from LSP), and

•

internal vehicles transferred (between cooperating DCs).

Furthermore, these measures were put in relation to the number of tours, number of customers, and shipped quantity. Finally, some additional analysis was done concerning the following output:

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer •

Percentage of vehicles from the internal fleets that remained unused,

•

percentage of vehicles used that were retrieved from the LSP,

•

percentage of vehicles used that were retrieved from another DC,

•

percentage of total capacity used,

•

percentage of capacity utilization of the vehicles used,

•

weight of vehicle costs in the total costs (in percentage), and

•

weight of external vehicle costs in the total costs (in percentage).

229

All of these measures were evaluated for each DC as well as for the supply chain as a whole. The purpose behind analyzing both the individual and the common results is to be able to depict if in a possibly beneficial supply chain result there are only winners or maybe there are winners and losers. In the former case, it is probably not necessary to perform any additional incentive activities among the participating companies in order to realize the cooperative setting. In the latter case, those companies that only “sacrifice” themselves have to be given some kind of motivation if the setting is to be realized. The following sections show, in the opinion of the author, the most interesting results. The complete results can be seen in Appendix E.5.

5.2.2.1 Total Costs Figure 83 gives an overview of the resulting total supply chain costs in which it becomes clear that the centralized planning dominates the other two strategies for any combination of settings. Also, the decentralized, cooperative planning dominates the decentralized, non-cooperative planning for any given setting. Though, in contrast to the centralized cooperative degree which is always the better choice, both decentralized options perform equally in five out of nine settings. In the other four, the decentralized, cooperative degree performs better. One first conclusion is, thus, that if in a supply chain the setting (concerning the categories shown here) is likely to change, only a centralized cooperation will ensure the best results. For example, in a setting in which the external vehicle price is half as much (on average) as the costs of internal transfer, a decentralized cooperation will not bring any benefits at all. It will just probably increase other costs not contemplated in this study such as transaction-related costs.

230

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Figure 83: Total Supply Chain Costs

The numbers shown in Figure 83 are average values of the ten experiments of each sample case. The ranges of the experiment results are depicted in the following figures:125 3500.00 2980.40

3000.00

Cost Units

2500.00 2000.00

2746.40 2815.93 2617.03 2506.40 2578.37 2552.79 2430.44 2396.79 2405.18 2186.83 2248.83 2210.63 2251.65 2251.65 2279.63 2155.77 2074.03 1989.86 1981.78 1989.86 1959.61 2021.61 1821.39 1821.39 1799.78 1861.78

1500.00 1000.00 500.00 0.00 .5 -1 .5 -1 Ce

.5 -1 .0 -1 Ce

.5 -1 .5 -0 Ce

.0 -1 .5 -1 Ce

.0 -1 .0 -1 Ce

.0 -1 .5 -0 Ce

.5 -0 .5 -1 Ce

.5 -0 .0 -1 Ce

.5 -0 .5 -0 Ce

Centralized, Cooperative Planning

Figure 84: Results of the Centralized, Cooperative Experiments in Terms of Total Supply Chain Costs

125

The experiments are denominated after the coding convention presented in Figure 82. For example, the sample space Ce-1.5-1.0 describes the experiments of centralized cooperation where the demand/capacity ratio is 1.5 (50 percent more demand than transportation capacity) and the external/internal vehicle price is 1.0 (the fixed price for a vehicle of the LSP is equal to the average vehicle transfer costs between DCs).

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

231

4500.00 3983.46

4000.00 3500.00

Cost Units

3000.00 2500.00

3749.46

3718.72 3519.82 3342.59 3315.82 3238.52 3186.59 2939.52 3001.52 3026.59 2961.13 3121.52 2961.13 2976.22 2882.62 2706.49 2781.39 2708.65 2715.02 2708.65 2793.02 2575.02 2575.02 2575.02 2635.02 3509.46

2000.00 1500.00 1000.00 500.00 0.00

.5 -1 .5 -1 Co De .5 -1 .0 -1 Co De .5 -1 .5 -0 Co De .0 -1 .5 -1 Co De .0 -1 .0 -1 Co De .0 -1 .5 -0 Co De .5 -0 .5 -1 Co De .5 -0 .0 -1 Co De .5 -0 .5 -0 Co De Decentralized, Cooperative Planning

Figure 85: Results of the Decentralized, Cooperative Experiments in Terms of Total Supply Chain Costs 4500.00 3983.46

4000.00 3500.00

Cost Units

3000.00 2500.00

3749.46 3509.46

3718.72

3519.82

3315.82 3186.59 2939.52 3001.52 3026.59 2979.52 3121.52 2885.39 2706.49 2781.39 2710.49 2715.02 2575.02 2575.02 2635.02

3238.52 3018.52 2986.79 2714.39 2793.02 2575.02

3342.59

2000.00 1500.00 1000.00 500.00 0.00 .5 -1 .5 -1 De

.5 -1 .0 -1 De

.5 -1 .5 -0 De

.0 -1 .5 -1 De

.0 -1 .0 -1 De

.0 -1 .5 -0 De

.5 -0 .5 -1 De

.5 -0 .0 -1 De

.5 -0 .5 -0 De

Decentralized, Non-Cooperative Planning

Figure 86: Results of the Decentralized, Non-Cooperative Experiments in Terms of Total Supply Chain Costs

The values in Figure 84, Figure 85, and Figure 86 show the maximum, minimum, and average results of each setting. It can be seen that an increase of demand is always paired with increasing costs in all three cooperative strategies. In contrast, an increase in vehicle price from the LSP is always paired with increasing costs only in the decentralized, non-cooperative environment. In the other two cooperative strategies there is one case each which does not affect the result quality (“Ce-0.5-1.0” to “Ce-0.5-1.5” and “DeCo-0.5-1.0” to “DeCo-0.5-1.5”). It becomes, thus, clear that obviously decentralized, non-cooperative planning results in a higher dependency from the pricing strategy of the external LSP than the other cooperative strategies.

232

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

The degree to which this dependency becomes reflected in the costing can be seen in the next section. In relative terms, the performance gap between the centralized and the decentralized cooperation is higher than the performance gap between the latter and the decentralized, non-cooperative settings as can be seen in the following tables:126 Table 37: Performance of Experiments (in Percentage) with an External/Internal Vehicle Price Ratio of 0.5 Retailer demand/transportation capacity ratio Cooperation degree

0.5

1.0

1.5

Centralized cooperation

100%

100%

100%

Decentralized cooperation 138.11%

137.58%

137.86%

No cooperation

137.58%

137.86%

138.11%

Table 38: Performance of Experiments (in Percentage) with an External/Internal Vehicle Price Ratio of 1.0 Retailer demand/transportation capacity ratio Cooperation degree

0.5

1.0

1.5

Centralized cooperation

100%

100%

100%

Decentralized cooperation 136.12%

133.72%

134.50%

No cooperation

133.84%

134.50%

136.22%

Table 39: Performance of Experiments (in Percentage) with an External/Internal Vehicle Price Ratio of 1.5 Retailer demand/transportation capacity ratio Cooperation degree

0.5

Centralized cooperation

100%

1.0

1.5

100%

100%

Decentralized cooperation 136.12%

130.56%

132.06%

No Cooperation

131.02%

132.06%

136.41%

It is interesting to note that the comparative performance gap in settings with an external/internal vehicle price ratio of 0.5 (see Table 37) is higher than in the other settings. In addition, the relative performance gap in settings with a demand/capacity ratio of 0.5 is, relatively speaking, the highest (see the first column in Table 37, Table 38, and Table 39). The conclusions

126

Table 37, Table 38, and Table 39 show the results of each setting in percentage. In each table, the centralized result sets the dimension of the comparison (i.e. it is 100% in each table). The values of the other cooperation degrees are calculated based on the centralized value of the corresponding column.

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

233

are, on the one hand, that with this model it seems the cheaper the external vehicle price is, the more beneficial it is to cooperate centrally. On the other hand, the more transportation capacity is available, the more beneficial it is to cooperate centrally. The explanation for this fact is that in scenarios with short capacities, the need for external vehicles in cooperative scenarios tends to be higher and, thus, one advantage of cooperation (being able to exchange vehicles if beneficial) can be applied less often (see also Section 5.2.2.2). Also, the results of the study show that the shorter the available capacity is, the higher the relative proportion of vehicle costs in the total costs is (i.e. the distance per vehicle/tour is lower; see Appendix E.5). This means the main advantage of the centralized cooperation (the reduction of distance-dependent variable costs) becomes less advantageous than in settings with more relative capacity (for a graphic visualization of these considerations, see also Figure 55, Figure 57, and Figure 59). Concerning the performance of the planning within a cooperative strategy, the results show that the worsening of the total costs based on both demand/capacity ratio and external/internal vehicle costs is higher within a centralized cooperation than in the other cooperative environments:127 Table 40: Performance of Experiments (in Percentage) in a Centralized Cooperation Retailer demand/transportation capacity ratio External/internal 0.5 vehicle cost ratio

1.0

1.5

0.5

100%

103.16%

122.74%

1.0

101.54%

110.01%

133.55%

1.5

101.54%

116.33%

143.70%

Table 41: Performance of Experiments (in Percentage) in a Decentralized Cooperation Retailer demand/transportation capacity ratio External/internal 0.5 vehicle cost ratio

1.0

1.5

0.5

100%

102.77%

122.51%

1.0

100.08%

106.51%

130.05%

1.5

100.08%

109.97%

137.40%

127

Table 40, Table 41, and Table 42 show the results of each setting in percentage. In each table, the lowest value is represented as the dimension of the comparison (i.e. 100% in each table). The values of the other cells in each table are calculated in relation to this value.

234

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Table 42: Performance of Experiments (in Percentage) with No Cooperation Retailer demand/transportation capacity ratio External/internal 0.5 vehicle cost ratio

1.0

1.5

0.5

100%

102.77%

122.51%

1.0

100.15%

106.61%

130.05%

1.5

100.29%

110.36%

137.40%

It can be seen that in the settings with over dimensioned capacities (first column in each table) the increase of the price for external vehicles has very little influence on the overall performance while in the settings with short capacity (third column in each table) the increase in the LDL vehicle price has a much larger impact on the quality of the result. Thus, it becomes obvious that the shorter the fleet of the DCs is dimensioned, the higher is the impact of LDL vehicle price since the need for external vehicles is comparably higher than if fleets are over dimensioned. Although the relative performance in a centralized, cooperative environment is increasingly worse in terms of absolute numbers for both ratios (see Table 40), it remains the dominant strategy for all settings described within this study. When taking a look at the individual results the situation reveals a slightly different picture. The obviously optimal strategy (provided by the assumptions made for this model; see Section 5.2), i.e. the centralization of planning, is not a win-win situation. While DC2 and DC3 perform clearly better, DC1 becomes the losing party in the cooperation in terms of total costs: 128 Table 43: Individual Performance of Experiments (in %) with an External/Internal Vehicle Price Ratio of 0.5 Retailer demand/transportation capacity ratio 0.5

1.0

1.5

Cooperation degree

DC1

DC2

DC3

DC1

DC2

DC3

DC1

DC2

DC3

Centralized cooperation

100%

100%

100%

100%

100%

100%

100%

100%

100%

Decentralized cooperation

91.46%

236.10% 156.04% 90.24%

237.50% 160.78% 90.69%

222.07% 177.53%

No cooperation

91.46%

236.10% 156.04% 90.24%

237.50% 160.78% 90.69%

222.07% 177.53%

128

Table 43, Table 44, and Table 45 show the individual results of each setting in percentage. In each table, the centralized result sets the dimension of the comparison (i.e. it is 100% in each table). The values of the other cooperation degrees are calculated based on the centralized value of the corresponding column.

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

235

Table 44: Individual Performance of Experiments (in %) with an External/Internal Vehicle Price Ratio of 1.0 Retailer demand/transportation capacity ratio 0.5

1.0

1.5

Cooperation degree

DC1

DC2

DC3

DC1

DC2

DC3

DC1

DC2

DC3

Centralized cooperation

100%

100%

100%

100%

100%

100%

100%

100%

100%

Decentralized cooperation

89.04%

236.10% 156.04% 85.23%

239.57% 166.81% 86.30%

228.88% 181.15%

No cooperation

89.22%

236.10% 156.04% 85.23%

239.57% 167.31% 86.30%

228.88% 181.15%

Table 45: Individual Performance of Experiments (in %) with an External/Internal Vehicle Price Ratio of 1.5 Retailer demand/transportation capacity ratio 0.5

1.0

1.5

Cooperation Degree

DC1

DC2

DC3

DC1

DC2

DC3

DC1

DC2

DC3

Centralized cooperation

100%

100%

100%

100%

100%

100%

100%

100%

100%

Decentralized cooperation

89.04%

236.10% 156.04% 81.72%

241.59% 171.77% 82.94%

235.53% 186.57%

No cooperation

89.59%

236.10% 156.04% 81.72%

241.59% 173.68% 82.94%

235.53% 186.57%

These tables show that DC1 performs best in the decentralized, cooperative settings while it performs worst at all centralized, cooperative settings (for absolute values, see Appendix E.5). In contrast, DC2 and DC3 both perform clearly better in the centralized than in the other scenarios. This is caused by the fact that in centralized settings, due to the Voronoi heuristics (see Section 5.1.2.1), DC1 takes over the supply of six retailers while DC2 and DC3 have to supply only three retailers each (in decentralized settings each DC distributes goods to four retailers each). This causes an overall reduction of distances and, thus, of variable costs; but it also means that DC1 must plan more tours and include more vehicles (for example, an average of 4.4 tours in experiments of the “Ce-0.5-0.5” sample space, compared to 2.5 tours in experiments of the “De0.5-0.5” sample space) and also cover a higher total distance (for example, an average of 944.07 distance units in experiments of the “Ce-0.5-0.5” sample space, compared to 912.75 distance units in experiments of the “De-0.5-0.5” sample space). On the other hand, as can be seen in Table 43, Table 44, and Table 45, DC2 and DC3 experience important cost reductions in a centralized environment compared to decentralized settings. For example, while DC2 must lead an average of 3 tours in experiments of the “De-0.5-0.5” sample space (covering an average of

236

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

875.55 distance units), it has to perform only an average of 1.8 tours in experiments of the “Ce0.5-0.5” sample space (covering an average of 364.48 distance units). In analogy, DC3 leads an average of 2.9 tours in experiments of the “De-0.5-0.5” sample space (covering an average of 830.57 distance units) compared to 1.9 tours in experiments of the “Ce-0.5-0.5” sample space (covering an average of 531.87 distance units). These figures point out the importance of the team assumption for the realization of centralized planning settings. When putting such a scenario into practice if a supply chain is a nonhierarchical environment (see Figure 1), DC2 and DC3 will be forced to create a corresponding incentive structure in order to motivate DC1 to cooperate. If incentives are not developed, the dominant strategy of DC1 is to take part in a decentralized cooperation.

5.2.2.2 Vehicle Costs When isolating the vehicle costs (i.e. the costs that are incurred before the vehicle covers any tour distance) from the distance-dependent variable costs, the study reveals some interesting figures:129 The vehicle costs comprise of fixed vehicle costs at each DC, transfer costs between DCs, and the price for vehicles of the LSP. These figures clearly show that centralized cooperation leads to the highest vehicle costs at any setting for this model. This means that, in this model, increasing vehicle costs (meaning fixed vehicle costs, transfer costs, and price of LSP vehicles used) lead to a lower performance of the centralized planning while lower vehicle costs increase the quality of a centralization of planning. This conclusion is corroborated by the results of analyzing the portion of vehicle costs in total costs (see Appendix E.5). Here, the results of centralized settings always have the highest proportion of vehicle costs in the total costs. Which means that in case of an average increase of vehicle costs, the impact in the quality of the centralized results is comparably larger than the impact on the quality of the decentralized planning options (for example, vehicle costs represent an average of 24.18% of the total costs in “Ce-1.5-1.5” experiments while in both “DeCo-1.5-1.5” and in “De-1.5-1.5” they represent 17.84%). In absolute figures, the cause of the higher vehicle costs in centralized settings can be found in the fact that after the reassignment of customers, DC1 takes over the supply of six instead of four retailers. This results in an increased need of additional vehicles which are retrieved either from the external LSP or from other DCs. In either case, this unequal distribution of retailers among DCs leads to more external retrieval of vehicles which is always more expensive than the

129

Table 46, Table 47, and Table 48 show the results of each setting in percentage. In each table, the centralized result sets the dimension of the comparison (i.e. it is 100% in each table). The values of the other cooperation degrees are calculated based on the centralized value of the corresponding column.

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

237

use of vehicles from one’s own fleet. For example, “Ce-1.0-1.0” experiments lead to an average of 8.1 tours that require an average of 3.4 additional vehicles (either from the LSP or from cooperating DCs), while both “DeCo-1.0-1.0” and “De-1.0-1.0” experiments lead to an average of 8.4 tours but only require an average of 2.6 additional vehicles. Table 46: Performance of Experiments in Terms of Vehicle Costs (in Percentage) with an External/Internal Vehicle Price Ratio of 0.5 Retailer demand/transportation capacity ratio Cooperation degree

0.5

Centralized cooperation

100%

1.0

1.5

100%

100%

Decentralized cooperation 73.91%

89.96%

97.01%

No cooperation

89.96%

97.01%

73.91%

Table 47: Performance of Experiments in Terms of Vehicle Costs (in Percentage) with an External/Internal Vehicle Price Ratio of 1.0 Retailer demand/transportation capacity ratio Cooperation degree

0.5

1.0

1.5

Centralized cooperation

100%

100%

100%

Decentralized cooperation 60.40%

83.79%

96.70%

No cooperation

84.66%

96.70%

61.63%

Table 48: Performance of Experiments in Terms of Vehicle Costs (in Percentage) with an External/Internal Vehicle Price Ratio of 1.5 Retailer demand/transportation capacity ratio Cooperation degree

0.5

Centralized cooperation

100%

1.0

1.5

100%

100%

Decentralized cooperation 60.40%

81.47%

97.66%

No cooperation

83.88%

97.66%

64.24%

Since in the centralized settings DC1 takes over the supply of additional retailers, it is not surprising that it carries the main load of these additional costs. While in both “DeCo-1.0-1.0” and “De-1.0-1.0” experiments DC1 requires 1.5 of the total 2.6 additional vehicles, in “Ce-1.01.0” experiments DC1 is the only DC that retrieves external vehicles (i.e. it requires the total 3.4 averaged external vehicles). Again, it becomes clear why in this model DC1 is the “losing party” in a centralized cooperation. Furthermore, the cause of the higher proportional part of vehicle costs in centralized experiments does not result from an increased amount of planned tours; in fact, it is the shorter distance per

238

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

tour that significantly reduces variable costs compared to the decentralized settings (see Section 5.2.2.3). As Table 46, Table 47, and Table 48 show, the quality gap (in terms of vehicle costs) between centralized and decentralized settings is reduced with increasing retailer demand (or decreasing transportation capacity). This is a result of the increasing need of LSP vehicles in all settings regardless if customers are reassigned or not which partly neutralizes the unequal distribution of retailers.

5.2.2.3 Distance Costs When considering distance-dependent variable costs (i.e. distances) to be isolated, the results draw a different picture. Since the main advantage of a centralized cooperation is the reassignment of retailers to the nearest DC, it can be expected to have a significant reduction of total distances in this scenario compared to decentralized ones. In addition, since the only difference between a decentralized cooperation and a non-cooperative environment is the potential exchange of vehicles (i.e. a potential reduction of vehicle costs), it can be expected that distances are the same in both decentralized settings. The next table shows the results in terms of distance costs:130 Table 49: Performance of Experiments in Terms of Distance-Dependent Variable Costs for all Settings of External/Internal Vehicle Price Ratio Retailer demand/transportation capacity ratio Cooperation degree

0.5

Centralized cooperation

1.0

1.5

1840.41

1840.41

2134.18

Decentralized cooperation 2618.39

2618.39

3052.92

No cooperation

2618.39

3052.92

2618.39

The results corroborate the expectations and show a significant increase (between 42.27 and 43.05 percent) of distances in decentralized settings. In addition, while settings with a demand/capacity ratio of 0.5 and 1.0 perform the same in terms of total distance when capacity becomes comparably short (i.e. demand/capacity ratio of 1.5), the total covered distance grows significantly by 15.96% in centralized settings and 16.60% in decentralized settings (see Appendix E.5).

130

Since the external/internal vehicle price has no influence on distances at all, Table 49 applies for all settings of external/internal vehicle price.

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

239

The reason behind these figures lies in the number of tours that have to be conducted in relation to the demand/capacity ratio. With increasing demand (by a given number of retailers), the amount of orders per vehicle is reduced due to the constant vehicle capacity of 500 load units. Thus, it becomes more difficult to satisfy several retailer orders in one tour:131 Table 50: Performance of Experiments in Terms of Total Tours for All Settings of External/Internal Vehicle Price Ratio Retailer demand/transportation capacity ratio Cooperation degree

0.5

Centralized cooperation

1.0

1.5

8.10

8.10

11.10

Decentralized cooperation 8.40

8.40

11.10

No cooperation

8.40

11.10

8.40

It becomes clear in Table 50 that the satisfaction of demand in settings with short (initial) capacities implies a higher number of vehicles. The tours that are conducted in these settings are more likely to be characterized by the supply of a single retailer in the context of direct tours (DC-retailer-DC): Table 51: Performance of Experiments in Terms of Distance per Tour for All Settings of External/Internal Vehicle Price Ratio Retailer demand/transportation capacity ratio Cooperation degree

0.5

1.0

1.5

Centralized cooperation

228.02

228.02

192.46

Decentralized cooperation 312.32

312.32

275.35

No cooperation

312.32

275.35

312.32

As Table 51shows, the distance per tour in settings with short capacity is lower than in the other settings since, of course, the covered distance in a direct delivery is always equal or shorter than the covered distance of a multi-stop tour. Table 51 shows also that decentralized planning always leads to shorter tours than both decentralized settings. This can have twofold causes: On one hand, it can mean that through the reassignment of retailers to the nearest DC, total distances are reduced. On the other hand, it can mean that through the unequal distribution of retailers, the capacity utilization of vehicles becomes less effective and centralized planning leads to less

131

Since satisfaction of retailer’s demand is mandatory in this model (see Section 5.1.2.1), the external/internal vehicle price has no influence on the decision wether a tour is to be conducted or not. Thus, Table 50 applies for all settings of external/internal vehicle price.

240

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

retailers per tour than decentralized settings.132 In order to address this aspect, the following table shows the retailers per tour. Table 52: Performance of Experiments in Terms of Retailers per Tour for All Settings of External/Internal Vehicle Price Ratio Retailer demand/transportation capacity ratio Cooperation degree

0.5

Centralized cooperation

1.0

1.5

1.49

1.49

1.08

Decentralized cooperation 1.43

1.43

1.08

No cooperation

1.43

1.08

1.43

As Table 52 shows, all settings with a demand/capacity ratio of 1.5 deliver on average fewer retailers than in the other settings. It is interesting to note that while in settings with sufficient transportation capacity (first two columns in Table 52) centralized settings are able to satisfy more retailer’s demand than decentralized (1.49 compared to 1.43 retailers per tour), in settings with short capacity (third column in Table 52) centralized planning leads to no better vehicle utilization than decentralized planning. Thus, it can be stated that the reduced distances in centralized settings with short capacities are merely a result of the reassignment of customers to the nearest DC. On the other hand, in settings with sufficient capacity centralized planning reduces distances due to both the reassignment of retailers to the nearest DC and a better vehicle capacity utilization. To which extent each of these two causes affect the results in terms of distances cannot be answered here. This is an aspect that ought to be analyzed in further simulations.

5.2.2.4 Capacities Following the argumentation of the preceding section, if one takes a look at the utilization of vehicle capacities (of those vehicles used to conduct a tour), then the results of the different sample spaces show interesting effects. As Table 53 shows, decentralized settings are not able to utilize the capacity of vehicles as effectively as centralized settings when total capacity is sufficient. The consequence is that in a decentralized planning environment there will be the tendency to have bigger vehicle fleets (or larger dimensioned vehicles) than in centralized environments.

132

It is thinkable that the reassignment could for example lead to the reassignment of those retailers with the highest demand to a single DC, which would result in this DC having to exclusively conduct direct deliveries.

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

241

Table 53: Performance of Experiments in Terms of Vehicle Capacity Utilization (in Percent) for All Settings of External/Internal Vehicle Price Ratio Retailer demand/transportation capacity ratio Cooperation degree

0.5

1.0

1.5

Centralized cooperation

74.07%

74.07%

81.08%

Decentralized cooperation 71.43%

71.43%

81.08%

No cooperation

71.43%

81.08%

71.43%

In contrast to this, when capacities are short the planning degree does not seem to have an impact on the capacity utilization at all (see third column in Table 53). This is probably due to the fact that single orders are larger and, thus, more tours are needed to deliver a single retailer (see Section 5.2.2.3). The consequence here is, thus, that the larger the total transportation capacity (assumption of the author: especially the vehicle capacity) is, the more beneficial it is to plan centrally. Thus, in supply chains planning decentralized, vehicle fleets will tend to be larger (in terms of number of vehicles, or in terms of vehicle dimensions, or both) than in supply chains planning the distribution centrally. In addition, with a given fleet dimension, supply chains with short capacities will benefit less from a centralized distribution planning than supply chains with sufficient transportation capacity. It is noticeable here that decentralized cooperation leads to no improvement in terms of capacity utilization. When taking a look at the utilization of the total transportation capacity (i.e. when including unused vehicles in the analysis), the results show a different picture. The following tables indicate the number of vehicles in the initial fleets of the DCs that remain unused after planning: Table 54: Performance of Experiments in Terms of Unused Internal Vehicles with a Retailer Demand/Transportation Capacity Ratio of 0.5 External vehicle price/avg. vehicle transfer costs Cooperation degree

0.5

Centralized cooperation

1.0

1.5

5.30

3.90

3.90

Decentralized cooperation 3.70

3.60

3.60

No cooperation

3.70

3.70

3.70

242

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Table 55: Performance of Experiments in Terms of Unused Internal Vehicles with a Retailer Demand/Transportation Capacity Ratio of 1.0 External vehicle price/avg. vehicle transfer costs Cooperation degree

0.5

1.0

1.5

Centralized cooperation

1.30

1.20

0.00

Decentralized cooperation 0.20

0.00

0.00

No cooperation

0.20

0.20

0.20

Table 56: Performance of Experiments in Terms of Unused Internal Vehicles with a Retailer Demand/Transportation Capacity Ratio of 1.5 External vehicle price/avg. vehicle transfer costs Cooperation degree

0.5

Centralized cooperation

1.0

1.5

0.30

0.00

0.00

Decentralized cooperation 0.00

0.00

0.00

No cooperation

0.00

0.00

0.00

These figures show that decentralized cooperation always needs the same number or more vehicles from the available fleets than in the non-cooperative setting. This is an expected result because the advantage of decentralized cooperation has the possibility of retrieving vehicles from other DCs which potentially reduces the final amount of unused vehicles in the supply chain. In addition, Table 54, Table 55, and Table 56 show that, except for one setting (see the third column in Table 55), centralized planning seems to need the same number or fewer vehicles than the decentralized settings (at least those available initially from the internal fleets). This fact can mean that in centralized cooperation either vehicles are being used more effectively (i.e. fewer tours are conducted), or more LSP vehicles are being retrieved, or both. The latter is confirmed by the figures shown in Table 50, Table 52, and Table 53. In order to corroborate the latter, it is necessary to look at the analysis of the retrieval of external vehicles (see Table 57, Table 58, and Table 59). Obviously, Table 54, Table 55, and Table 56 show that the amount of unused vehicles decreases with increasing demand/capacity ratio. Nevertheless, Table 56 reveals that when capacities are too short, the centralized planning still does not make use of all available vehicles. In contrast, decentralized settings do not leave any vehicles from internal fleets unused. Since the number of tours in centralized situations is larger than the number of internal vehicles available, it is evident that centralized planning retrieves more LSP vehicles than decentralized settings even if internal vehicles are still available.

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

243

Table 57: Performance of Experiments in Terms of LSP Vehicles Retrieved with a Retailer Demand/Transportation Capacity Ratio of 0.5 External vehicle price/avg. vehicle transfer costs Cooperation degree

0.5

Centralized cooperation

1.40 0.00 more LSP vehicles more effective

1.0

1.5 0.00 more effective

Decentralized cooperation 0.10

0.00

0.00

No cooperation

0.10

0.10

0.10

Table 58: Performance of Experiments in Terms of LSP Vehicles Retrieved with a Retailer Demand/Transportation Capacity Ratio of 1.0 External vehicle price/avg. vehicle transfer costs Cooperation degree

0.5

1.0

1.5

Centralized cooperation

3.40 more effective more LSP vehicles (0.8 LSP vehicles more, but 1.1 vehicles less used)

3.30 2.10 more effective more effective more LSP vehicles (0.9 LSP vehicles more, but 1.1 vehicles less used)

Decentralized cooperation 2.60

2.40

2.40

No cooperation

2.60

2.60

2.60

Table 59: Performance of Experiments in Terms of LSP Vehicles Retrieved with a Retailer Demand/Transportation Capacity Ratio of 1.5 External vehicle price/avg. vehicle transfer costs Cooperation degree

0.5

Centralized cooperation

5.40 5.10 more LSP vehicles

1.0

1.5 5.10

Decentralized cooperation 5.10

5.10

5.10

No cooperation

5.10

5.10

5.10

These tables reveal that, except for one setting again (see third column in Table 58), centralized planning leads to more or equal retrieval of external vehicles than decentralized settings. The cause can be found in the fact that the reassignment of customers leads to a higher need of additional vehicles by DC1. Since the source of additional vehicles depends on the distance between DCs and also the LSP price, sometimes it is more cost effective to leave internal vehicles from other DCs unused. If the values in these tables are compared with the figures in Table 54, Table 55, and Table 56, one can see that while decentralized cooperation is never more effective (in terms of vehicle capacity utilization), there are three settings (characterized with “more effective”) where

244

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

centralized planning leads to a better utilization of vehicle capacity and, thus, to fewer vehicles used. In addition, there are two settings (characterized by “more LSP vehicles”) in which the higher amount of unused internal vehicles in centralized cooperation is justified only by the higher amount of LSP vehicles retrieved. Furthermore, there are two settings (see Table 55) in which the higher number of unused internal vehicles in central cooperation is caused by both retrieval of more LSP vehicles and a better capacity utilization. Finally, it can be seen that in the case of no-cooperation, all cases which use fewer internal vehicles are due to the inability to retrieve vehicles from other DCs and not because of a better capacity utilization. In terms of capacity utilization, both decentralized settings are equal. It is interesting to note that in five out of six settings with sufficient transportation capacity (see Table 54, Table 55), centralized cooperation leads to a more effective capacity utilization while short capacities neutralize this effect completely (see Table 56). This means that if capacities are not understood to be sunk costs (i.e. capacities are subject of redimensioning for future periods) in supply chains with enough transportation capacity, centralized planning allows one to redimension the fleets to a higher extent than decentralized planning. On the other hand, the fact that centralized planning requires more external vehicles leads to the conclusion that due to the unequal distribution of retailers, a repositioning of vehicles among DCs leads to even better results. Of course, the assumption here is that the decision whether to cooperate or not is a repetitive action in every period and the characteristics of the planning stay the same. If this is not the case, centralized cooperation generally leads to a higher dependency on the external LSP than decentralized settings (except for “Ce-1.0-1.5” settings, see third column in Table 55). If the dependency from external LSPs is to be avoided, decentralized cooperation seems to be the best strategy (again, except for “DeCo-1.0-1.5” settings). In the opinion of the author, the effect that centralized planning tends to use more external vehicles is possibly the result of the geographical characteristics of the scenario. This is another parameter that could be analyzed in further simulations in which experiments would be based on variations of the centrality of actors needing additional vehicles in the plane. It has to be noted again that in centralized settings DC1 is the cooperating partner that carries most of the additional vehicle costs (if not all) compared to decentralized settings (see Appendix E.5).

5.2.2.5 Relative Results If one analyzes the results by putting them in relation to the amount of retailers, one can see that, for example, in centralized settings DC1 not only faces the supply of additional customers but

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

245

also faces more demand per customer than in decentralized settings. It can be assumed that the “losing part” played by DC1 in centralized cooperation would not be that accentuated (see Table 43, Table 44, and Table 45) if the reassignment would not increase the per customer demand for DC1. To confirm this assumption, further simulations have to be run in which the per retailer demand supplied by individual DCs is differentiated. In this study, DC1 always faces an increase of demand per retailer between 9.96% and 22.74% while DC2 and DC3 face the corresponding reduction between 8.96% and 16.13% (for DC2) and between 1.48% and 12.92% (for DC3). In contrast, although in absolute numbers DC1 faces higher costs (both vehicle and distancedependent costs) as well as higher loads per customer, the costs per customer show that in centralized settings not only DC2 and DC3 face lower costs per customer but also DC1 has significant reductions: Table 60: Individual Performance of Experiments in Terms of Total Costs per Customer (in Percentage) with an External/Internal Vehicle Price Ratio of 0.5 Retailer demand/transportation capacity ratio 0.5

1.0

1.5

Cooperation degree

DC1

DC2

DC3

DC1

DC2

DC3

DC1

DC2

DC3

Centralized cooperation

170.44

128.69

183.62

180.78

128.69

183.62

223.25

156.48

198.74

Decentralized cooperation

233.84

227.89

214.89

244.69

229.24

221.42

303.71

260.62

264.62

No cooperation

233.84

227.89

214.89

244.69

229.24

221.42

303.71

260.62

264.62

Table 61: Individual Performance of Experiments in Terms of Total Costs per Customer (in Percentage) with an External/Internal Vehicle Price Ratio of 1.0 Retailer demand/transportation capacity ratio 0.5

1.0

1.5

Cooperation degree

DC1

DC2

DC3

DC1

DC2

DC3

DC1

DC2

DC3

Centralized cooperation

175.49

128.69

183.62

203.14

128.69

183.62

253.92

156.48

208.02

Decentralized cooperation

234.38

227.89

214.89

259.69

231.24

229.72

328.71

268.62

282.62

No cooperation

234.84

227.89

214.89

259.69

231.24

230.42

328.71

268.62

282.62

246

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

Table 62: Individual Performance of Experiments in Terms of Total Costs per Customer (in Percentage) with an External/Internal Vehicle Price Ratio of 1.5 Retailer demand/transportation capacity ratio 0.5

1.0

1.5

Cooperation degree

DC1

DC2

DC3

DC1

DC2

DC3

DC1

DC2

DC3

Centralized cooperation

175.49

128.69

183.62

223.78

128.69

183.62

283.82

156.48

214.52

Decentralized cooperation

234.38

227.89

214.89

274.32

233.19

236.55

353.09

276.42

300.17

No cooperation

235.82

227.89

214.89

274.32

233.19

239.19

353.09

276.42

300.17

These figures show that in all settings, DC1 faces cost reductions per supplied retailer; although the total costs (see Table 43, Table 44, and Table 45) have increased. This means that probably, if retailers are to pay the same amount for the delivery of products after a centralization of distribution planning, DC1 will be able to at least neutralize the additional financial burden. Of course, the assumption here is that retailers are not part of the unity of the supply chain and, thus, do not take advantage of reduced costs at the preceding echelons. The margins at the DC stage, though, become larger and they experience, thus, a flexibility increase in terms of pricing. The cause for this reduction of costs per retailer at each DC lies in the distances covered to reach each retailer. Table 63: Performance of Experiments in Terms of Distance per Retailer for All Settings of External/Internal Vehicle Price Ratio Retailer demand/transportation capacity ratio Cooperation degree

0.5

Centralized cooperation

1.0

1.5

153.37

153.37

177.85

Decentralized cooperation 218.20

218.20

254.41

No cooperation

218.20

254.41

218.20

In all experiment settings, centralized planning reduces the covered distance per retailer significantly (between 43.27% and 43.05%). And DC1 experiences reductions of distance per retailer as well as DC2 and DC3.

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

247

These numbers are correlated by the analysis of costs per tour. In this case, DC1 becomes the losing party in centralized settings in terms of absolute numbers, and the costs per tour show a win-win situation for all involved DCs: Table 64: Performance of Experiments in Terms of Total Costs per Tour with a Retailer Demand/Transportation Capacity Ratio of 0.5 External vehicle price/avg. vehicle transfer costs Cooperation degree

0.5

1.0

1.5

Centralized cooperation

242.63

246.29

246.29

Decentralized cooperation 322.78

323.02

323.02

No cooperation

323.23

323.66

322.78

Table 65: Performance of Experiments in Terms of Total Costs per Tour with a Retailer Demand/Transportation Capacity Ratio of 1.0 External vehicle price/avg. vehicle transfer costs Cooperation degree

0.5

Centralized cooperation

1.0

1.5

250.32

266.79

281.98

Decentralized cooperation 331.69

343.68

354.73

No cooperation

344.03

356.05

331.69

Table 66: Performance of Experiments in Terms of Total Costs per Tour with a Retailer Demand/Transportation Capacity Ratio of 1.5 External vehicle price/avg. vehicle transfer costs Cooperation degree

0.5

1.0

1.5

Centralized cooperation

216.86

235.93

253.80

Decentralized cooperation 298.98

317.27

335.10

No cooperation

317.27

335.10

298.98

As can be seen in Table 64, Table 65, and Table 66, centralized planning leads to significant reductions in terms of total costs per tour while decentralized cooperation leads to small reductions. These reductions per tour apply to all three collaborating DCs including DC1 (see Appendix E.5). Here it is interesting to note that experiments with a demand/capacity ratio of 1.0 (see Table 65) have the highest costs per tour in this study while experiments with short capacity (see Table 66) have the lowest costs per tour. This has twofold causes: First, settings with a demand/capacity ratio of 1.0 and 0.5 (Table 64) conduct fewer tours and, thus, comparably longer tours. Second, the vehicle costs per tour are higher in experiments with a demand/capacity ratio of 1.0 than experiments with 0.5 because DCs have to retrieve more LSP

248

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

vehicles than in settings with a ratio of 0.5 and conduct the same amount of tours. Although experiments with a demand/capacity ratio of 1.5 face the highest vehicle costs per tour, this is neutralized by the fact that here the distance per tour (due to more direct deliveries) is lowest. Of course, the latter settings face an increased number of tours which leverages this effect in the total sum (see Table 50).

5.2.3 Summary of Results The bottom line of this study shows that in terms of total supply chain costs, centralized cooperation always ensures the best planning results while decentralized cooperation is always better or equal to non-cooperative settings. But as revealed in Section 5.2.2.1 while cooperation is beneficial for the supply chain as a whole, DC1 plays the “losing part” in the centralized setting. In contrast, in Section 5.2.2.5 it was shown that in terms of costs per retailer, all parties are winners in the centralized cooperation. Thus, if one analyzes the potential revenue of the cooperating parties based on the degree of cooperation, the results are as follows:133 Table 67: Individual Performance of Experiments in Terms of Potential Revenue Units with a Retailer Demand/Transportation Capacity Ratio of 0.5 External vehicle price/avg. vehicle transfer costs 0.5

1.0

1.5

Cooperation degree

DC1

DC2

DC3

DC1

DC2

DC3

DC1

DC2

DC3

Centralized cooperation

380.39

297.58

93.81

356.15

297.58

93.81

362.00

297.58

93.81

Decentralized cooperation

0.00

0.00

0.00

1.84

0.00

0.00

5.74

0.00

0.00

No cooperation

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

133

Potential revenues are calculated assuming that the costs per retailer in non-cooperative settings are always the price charged to each retailer by each DC that supplies it. Thus, in non-cooperative settings, the total revenue is always 0. In decentralized, cooperative settings, the revenue is calculated as the difference between the price charged and the costs per retailer of each decentralized, cooperative setting multiplied by the amount of retailers supplied by each DC (4 each). In centralized settings, the revenue is calculated as the difference between the price charged and the costs per retailer of each centralized, cooperative setting multiplied by the amount of retailers supplied by each DC (6 by DC1, 3 by DC2, and 3 by DC3).

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

249

Table 68: Individual Performance of Experiments in Terms of Potential Revenue Units with a Retailer Demand/Transportation Capacity Ratio of 1.0 External vehicle price/avg. vehicle transfer costs 0.5

1.0

1.5

Cooperation degree

DC1

DC2

DC3

DC1

DC2

DC3

DC1

DC2

DC3

Centralized cooperation

383.49

301.63

113.39

339.33

307.63

140.39

303.23

313.48

166.71

Decentralized cooperation

0.00

0.00

0.00

0.00

0.00

2.77

0.00

0.00

10.57

No cooperation

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Table 69: Individual Performance of Experiments in Terms of Potential Revenue Units with a Retailer Demand/Transportation Capacity Ratio of 1.5 External vehicle price/avg. vehicle transfer costs 0.5

1.0

1.5

Cooperation degree

DC1

DC2

DC3

DC1

DC2

DC3

DC1

DC2

DC3

Centralized cooperation

482.76

312.42

197.64

448.76

336.42

223.79

415.61

359.82

256.94

Decentralized cooperation

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

No cooperation

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

As Table 67, Table 68, and Table 69 show, in terms of potential revenues, all DCs profit from centralized cooperation at all settings, and DC1 and DC3 profit from decentralized cooperation in two settings each. DC2 never profits from decentralized cooperation. Thus, if customers in such a model (i.e. the retailers) are not part of the decision and do not take advantage of lower total costs per retailer, DCs will be motivated to take part in the centralized cooperation since they will be able to “charge” retailers the “decentralized” price and have a higher margin than before. This will neutralize the effects of increased absolute total costs for DC1 and enable a centralized cooperation without needing a team assumption. In contrast, if all cost advantages are passed through to the retailers, DCs will have no individual motivation to participate in cooperation which again requires a team assumption or some kind of incentive structure or hierarchical power to put cooperative strategies into practice.

250

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

The findings of this computational study show several isolated effects that influence the tendency to cooperate or not. For example, Section 5.2.2.1 shows that in supply chains with relatively low LSP prices, centralized planning achieves comparably better results than in settings with high LSP vehicle prices (see Table 38, Table 39, and Table 40). Thus, ceteris paribus, increasing external vehicle prices lead to less centralization. This is corroborated by the fact that fixed vehicle costs assume a higher portion of the total costs in centralized cooperative settings than in decentralized settings. Thus, ceteris paribus, increasing LSP costs have a higher relative impact in supply chains with centralized cooperation than in supply chains that perform the distribution planning decentralized. Here, decentralized cooperation suffers the lowest relative impact of increasing LSP costs. In absolute terms, of course, non-cooperative environments are more dependent on LSP pricing than others. Obviously, greater distances between DCs lead to less centralization since one of the advantages of cooperation is the potential exchange of vehicles between DCs which becomes more expensive with increasing distance. Another isolated effect shown in Section 5.2.2.1 is that, ceteris paribus, the more transportation capacity is available, the more beneficial it is to cooperate centrally. This is a consequence of the availability of unused vehicles to exchange among partners and the increased need of external vehicles which partly neutralizes the unequal distribution of retailers among DCs. For both cooperative environments, it can be observed that the relative gap between the quality of the results decreases with increasing demand (decreasing capacity) compared to non-cooperative settings. In terms of distance costs, it can be seen that probabilistic distances and distance-dependent costs (due to e.g. traffic jams, accidents, etc.) have a higher relative impact on decentralized settings than in centralized ones. Additionally, while in settings with short capacities the reduced distances in centralized cooperation are merely a result of the reassignment of customers to the nearest DC, in settings with sufficient capacity, centralized planning reduces the distances because of both the reassignment of retailers to the nearest DC and a better vehicle capacity utilization. The observable effect here is that the shorter the capacities are calculated, the higher the impact of distance-dependent variable costs is. This is probably more due to larger single orders compared to the vehicle capacity than a consequence of the number of vehicles. This fact leads to more direct deliveries and, thus, to higher total distances covered. In an analogy, it can be said that high variable costs will lead to higher dimensioned vehicles that will, thus, be able to deliver more retailers in one tour. The vehicle capacity in relation to the size of orders seems, thus, to be an interesting parameter to vary in further simulations.

5.2 An Exemplary Evaluation of Cooperative Planning Using the SCOptimizer

251

In general, as shown in Section 5.2.2.4 capacities are better used in central cooperative settings. Supply chains that do not plan centrally will tend to have larger vehicle fleets (or larger sized vehicles). And it is for those supply chains with larger transportation capacities for which a centralization is most beneficial. In this context, it cannot be affirmed to which extent this influence comes from the vehicle capacity or from the amount of available vehicles. This is a parameter that needs to be analyzed in further simulations. It can be affirmed, though, that the shorter the transportation capacities are calculated, the higher the impact of distance-dependent variable costs in the performance of the planning. In particular, the vehicle capacity seems to play an important role, but in order to determine to which extent a further simulation has to take place in which settings are to be calculated based on the vehicle capacity. Another interesting conclusion is the fact that, although the planning of the tours is always performed with the Savings method (regardless of the cooperation degree), the vehicle capacities seem to be used more effectively in centralized settings if capacities are sufficient (see Table 52). This can be due to the fact that in the centralized experiments one of the DCs is able to combine the supply of six retailers (compared to four) in order to determine the highest savings. This fact seems to have a greater impact on distances than the fact that the other two DCs have fewer retailers to combine (three each) in order to plan the tours. On the other hand, large retailer orders seem to neutralize this effect (see Table 52). An assumption that, thus, can be made but has to be confirmed by corresponding simulations is that if orders cannot be split, the larger single orders are, the less effective it is to plan tours centrally and, thus, the higher the tendency to decentralize the distribution planning. Correspondingly, the larger the vehicles are sized, the higher the tendency to plan centrally. In any case, if the cooperative decision is a repetitive game, ceteris paribus, centralization permits a reduction of vehicle fleets while decentralized cooperation leads to no improvement in capacity utilization at all. The conclusion in terms of capacities is that short-term available capacity in supply chains leads to less cooperation (both central and decentral). As a conclusion, it has to be noted critically that centralized planning as described here (see Section 2.3.4) is likely not to be realized in most supply chains. The reluctance to give up decision power is a plausible assumption that can probably not be broken with the mere prospect of benefits paired with a direct dependency of a third entity that is probably not free of subjectivity and individual interest. The game theory issues and the social implications of dependencies present challenges that let one assume it would be very difficult to implement centralized decision structures among independent actors in supply chains.

252

5. SCOptimizer – A Prototype for Quantifying Benefits of Cooperative Planning in Supply Chains

On the other hand, decentralized cooperation is a much more plausible scenario to become realized since there is no shift in decision power and there is a prospect of benefits (but not as high as in centralized settings).

6 Summary and Conclusions

This dissertation presented a prototypical approach of the quantification of the benefits of cooperative planning in supply chains. It addressed what seems to be an important reason for the reluctance of implementing SCM instruments such as SCM software solutions for cooperative planning. For this purpose, the dissertation first analyzed relevant theoretical frameworks (namely the Theory of Cooperation and the Theory of Supply Chain Management) in order to allow an operational differentiation of conceptually different cooperation degrees. Furthermore, the field of transportation, as one of the most important ones in terms of costs, was selected for deeper analysis. The literature was approached in order to identify existing cooperation scenarios that could serve to improve supply chain performance in existing real-life scenarios. Such a scenario is presented in the context of a case study analysis that focused on the bullwhip effect. This effect serves as an indicator of the lack of cooperation, or of the quality of the cooperative strategy, or of both. Both transportation and the bullwhip effect are exemplarily implemented in the prototype that allows a model-based quantification of the benefits of cooperation depending on the resulting costs. Since the formulation of general statements concerning the quality of planning cannot be based on a single experiment, a computational study with 270 experiments depicts the sensitivity of results in dependency of relevant planning parameters. Supply chains, such as the one described in the case study, mostly comprise of legally independent organizations that will likely not give up planning and decision power to a central entity to ensure best overall supply chain results. It is plausible to assume that decentralized, cooperative planning will have a higher acceptance among potential cooperation partners. As a conclusion, the first section of this chapter (Section 6.1) summarizes the results of the research efforts of this dissertation and the implications that can be derived from them. In addition, Section 6.2 provides a short outlook on further research work that, in the opinion of the author, complements the findings of the dissertation.

254

6. Summary and Conclusions

6.1 Summary of the Findings and Implications Considering the central research questions of this dissertation and the goals implied in these, the analysis of the theoretical framework allowed to define the scope of research: cooperation as core aspect of SCM and global optimization (i.e. centralized cooperative planning) as the ultimate goal of SCM strategies. In addition, both transportation as a main task of cooperative business logistics and the bullwhip effect as an indicator for the efficiency of cooperation in supply chains were selected to provide the focus of this dissertation. Eight exemplary scenarios taken from the literature provided the basis for the analysis of potential benefits of cooperation in terms of reduction of the bullwhip effect in the Audi A8 V8 4.0l diesel engine supply chain. The explorative survey presented in Chapter 3 showed that, for the given sample, the reluctance to implement SCM software as an instrument for SCM is mainly caused by the difficulty to determine the actual benefits that such a solution might bring to the organization. Thus, the quantification of the benefits of cooperation through the use of a SCM solution seems to be the biggest challenge in the European automotive industry when it comes to decide whether to perform ICT-based SCM or not. This fact provides first implications to SCM software vendors who, in the opinion of the author, ought to complement current sales efforts with additional simulations of the supply chain context of potential customers rather than changing pricing principles or providing additional technical background to the decision makers. SCM software seems to provide benefits to organizations (as revealed in the survey) but improvement in cooperation is considered to be less important than other goals. IN this context, companies rank the unclear cost/benefit-ratio as the major problem of an inter-organizational use for these software solutions. It seems thus that the main use of SCM software is not global but rather local optimization. The consequences for SCM software vendors can be twofold. SCM software should not be primarily developed to provide global optimization functionality to customers since they seem not to be willing to “surrender decision-making power to an unbiased decision maker” (SimchiLevi et al., 2002, p. 56). On the other hand, if SCM software is to become an instrument for realization of SCM and its main objective (i.e. centralized planning), SCM software vendors ought to provide potential customers with corresponding figures that reveal the context-driven costs and benefits of such cross-company implementations.

6.1 Summary of the Findings and Implications

255

Under the perspective of the central research questions of this dissertation it can be said that for this sample from the European automotive industry the quantification of the benefits of SCM software and of its inter-organizational use represent the biggest challenge for its adoption. In addition, the survey revealed the existence of indirect network effects in the SCM software market. This fact could possibly lead SCM software vendors with big market shares in complementary markets (such as ERP software market) to concentrate even more in the compatibility of SCM solutions to ERP solutions of their own rather than trying to develop generic interfaces with open standards. On the other hand, SCM software vendors with low or non-existing market share in complementary markets are provided with the evidence that the compatibility of their SCM software solution with leading ERP solutions is mandatory to subsist in the SCM software market. The understanding of SCM software as stand-alone solution seems not to be applicable. Also, the preparation of SCM software for best-of-breed strategies loses importance for a SCM software vendor with a big market share in the ERP software market while, for the vendor with a small or non-existing ERP software market share, it becomes of key relevance. The second empirical approach of this dissertation, the case study of Audi AG, discloses several interesting and, as far as the author can tell, new aspects concerning the causes of bullwhip effect in supply chains of the automotive industry. First, the case study revealed the existence of an important bullwhip effect that does not seem to be caused by the four major causes described in the literature. The firms involved in the analysis are providing their respective supplier with rolling call-offs, announcing demands in advance and agreeing to certain levels of commitment for current call-off orders, a procedure that is very common in the European automotive industry. Basically, this aims at reducing uncertainty and thus enhancing planning predictability. However, an opposite effect arises when the levels of commitment to the rolling call-off plans between echelons do not correspond and are not synchronized. Furthermore, while collecting data for the analysis, the author could observe some apparently irrational behavior from certain interviewees when trying to determine the causes for over dimensioned orders. The lack of cooperation hereby and the observations made during the case study lead the author to the conclusion that goods are disappearing at some stage of the supply chain between TCG Systemtechnik and Audi Hungaria. Under the perspective of the central research questions of this dissertation, it can be said that the supply chain of the Audi A8 V8 4.0l diesel engine is a cooperative one since partners provide supply call-offs that give a preview of the next six months of orders. Nevertheless, some inefficiency could be identified. First, the non-harmonic commitment to previews in supply call-

256

6. Summary and Conclusions

offs throughout the supply chain is an additional source of uncertainty that contributes to accentuate the bullwhip effect. Second, the lack of effective control that results in tolerated withdrawal of goods represents a further source of variability that enhances the bullwhip effect. The implications of these findings for decision makers are twofold. First, when agreeing on supply contracts a higher system thinking has to be present in order to synchronize levels of commitment to supply call-offs throughout the supply chain. This will clearly reduce uncertainty at some stages of the supply chain and thus reduce the bullwhip effect. Second, in order to achieve an efficient flow of goods throughout the supply chain, monitoring becomes a key element to ensure the consistency of order amounts, production quantities, and inventory levels. Corresponding ICT solutions and data integration for higher transparency are mandatory. Further, satellite teams that perform observation activities in the supply chain can also provide a more effective control of the flow of goods. As mentioned above, the quantification of the benefits of cooperation represents one of the key issues for adopting SCM software and SCM strategies. As a response to this, this dissertation presents the SCOptimizer prototype. This application is designed to combine conceptual degrees of cooperation (as identified in the theoretical analysis) and SCM planning methods (exemplarily distribution and order/demand planning within the bullwhip effect analysis) allowing comparison of results and thus the quantification in terms of model costs. This prototype uses open technologies and open standards to provide a state-of-the-art implementation that ensures modularity and thus flexibility, compatibility and extensibility of the solution. This allows maximum flexibility for modelling combined with an extensible repository of planning classes that allow the comparison of cooperation degrees. In this manner, planners and developers that follow the design patterns of this prototype are able to asses the benefits of planning activities that are not limited to a specific supply chain or a specific context but can be adapted to basically any real-life scenario to ensure an accurate quantification. XML plays the key role for the realization of this flexibility. All data and interfaces are xml based; all configuration patterns rely on XML and allow thus extensibility without additional coding within the SCOptimizer (merely the planning method has to be implemented). In order to give evidence of the applicability of the SCOptimizer for quantifying the benefits of different cooperative strategies the dissertation presented a computational study. This study addresses an exemplary distribution scenario and describes how the quantification might be performed in dependency of relevant planning parameters. The results show that for this distribution scenario “no cooperation” is a dominated strategy and mostly non-pareto optimal. This means that there is room for negotiations that will provide improvement for all partners. On

6.1 Summary of the Findings and Implications

257

the other hand, “centralized cooperation” brings the most benefits for the supply chain but also might cause individual partners to perform worse than in decentralized settings. In the theoretical approach the team assumption is often used to describe central settings. In reality, cooperation in supply chains is mostly characterized by the lack of hierarchical structures which revokes the team assumption and accentuates the importance of corresponding contract negotiations that open up the share of risks and benefits between partners. For the example shown in the computational study it remains of course unclear, if in practice DCs are open to giving up their bonds to customers. Nevertheless, the application of a quantification instrument such as the SCOptimizer allows decision makers to have additional information on which to base cooperation related decisions such as whether or not to implement SCM software, whether or not to perform centralized, cooperative planning, whether or not to participate in an EDI network for more effective exchange of information with partners, etc. In addition, the quantification can also serve as a basis for anticipated organizational changes for the adoption of cooperation since the computational study might ex-ante reveal over dimensioned capacities, among other relevant aspects.

6.2 Outlook and Further Research The results of the model-driven quantification with the SCOptimizer are, as already mentioned, an additional information that is to help decision makers to have a better base for defining and implementing cooperative strategies with partners in the supply chain. The SCOptimizer though has a set of implied assumptions that need to be revised by decision makers before coming to the conclusion whether or not to implement cooperation. On the one hand, the SCOptimizer does not consider communication costs. It is assumed that the communication processes involved in the decentralized and centralized, cooperative planning do not generate significant cost (i.e. communication costs are assumed to tend towards zero which might not be a farfetched assumption considering Internet based communication). On the other hand, the time factor for taking a decision respective creating a plan is neither taken into consideration as relevant factor for the quality of a plan. It can be assumed that due to the amount of communication processes involved in both decentralized and centralized, cooperative planning, the final decision (i.e. the valid plan) will probably take some more time than the decentralized, non-cooperative decision. In addition, the additional complexity of the planning in the centralized cooperation might often require longer computing times in order to create the valid plan. This plan will then have to be published among the cooperating partners and probably be suitable to enhancements, changes, or corrections by other actors.

258

6. Summary and Conclusions

The combination of the quality of a plan with the time factor and the consideration of the impact of communication on the final result is a task that can enhance the applicability of results to reallife scenarios (for an analysis of the impact of different decision structures within the context of an investment problem considering time and communication costs see Buxmann, 2001, pp. 120ff). In analogy, the possibility of making an accurate decision on whether or not to cooperate with partners can only be accomplished with a detailed analysis of costs involved in the cooperation. Transaction costs (see for example Rotering, 1993) and ideally all supply chain related costs (for example Stölzle & Otto, 2003; Zäpfel & Piekarz, 1996) have to be taken into account in order to be able to determine the added value of cooperation. Since its precise estimation remains a challenging task, this aspect must be tagged with a question mark where additional research is needed. Nevertheless, if no benefits are quantified when comparing planning results, no further costing and transactional analysis must be done in order to determine the lack of applicability of a certain cooperative degree. This dissertation draws upon a first-order model of benefits (Barua et al., 1995; Mukhopadhyay & Kekre, 2002). “First-order benefits are related to firm actions and can be influenced directly by firms. In contrast, second-order benefits are competitive outcomes and incorporate the influence of external factors such as competitors' moves and environmental changes that are beyond the control of an individual firm.” (Subramani, 2003, p. 12). The benefits quantified by the approach of the SCOptimizer are merely first-order benefits that relay on operative or strategic benefits. On the other hand, an accurate decision would require taking second-order benefits into consideration. This calls for an estimation of the development of the market after the strategic decision including the consideration of the moves of competitors in other supply chains and of competitors of the partners involved in the cooperation. Of course, such estimations probably entail simulation frameworks such as the ones needed for weather estimation. Nevertheless, further efforts are needed in this direction. Considering the operationalization of decision structures, currently the SCOptimizer implements the division presented by Wyner and Malone (1996). A further differentiation like the one presented by Schneeweiss (1999, p. 5; see also Figure 12) invites to a more specific quantification that takes a further conceptual division of cooperation forms. In future, such an implementation in the fashion of the SCOptimizer is likely to allow additional ex-ante analysis of cooperation. An interesting approach on the quantification of the added value of instruments to support cooperation is to combine the analysis of set-up and running costs with the quantification of the benefits achieved with its use. This would allow to asses the benefits of different degrees of

6.2 Outlook and Further Research

259

customization of, e.g., a SCM solution. More customization at implementation stage means both more set-up costs and probably a better support of specific business processes within the implementing organization or supply chain (i.e. less process-driven costs). In contrast, less customization leads to a worse performing support of specific processes and thus probably to higher process-dependent costs. With the aid of a tool like the SCOptimizer, the quantification of different levels of process support would lead to a better assessment of the ideal customization level of cooperation instruments like SCM software. Under the perspective of the explorative survey presented in Chapter 3, the question arises if the findings are representative for the automotive industry in general and if those results are also portable to other branches. Is the quantification of benefits of cooperation the main challenge for adoption in other branches? The case study calls for additional simulation efforts to determine how the bullwhip effect performs if commitment to supply call-offs are synchronized in the entire supply chain. Further case studies that analyze this aspect will have to follow in future to allow an accurate assessment of the impact of commitment to supply call-offs to the overall supply chain costs. The contribution of this dissertation to the field of SCM focuses primarily on the quantification of the benefits of cooperation. It was shown that a centralized cooperation is the dominant strategy in a supply chain centric model of competition. The efforts of the dissertation show that the global view of supply networks as an entity that competes with other supply chains offers synergies that allow to realize important cost savings and improvement of service levels. Still, questions remain: •

Who will be the central institution that plans for all partners?

•

How will risks and benefits be split between partners?

The answers consider and involve the design and negotiation of contracts, the existence of trust relationships, the imposing of decision to ‘weaker’ partners, the development of mathematical models for global optimization and benefits sharing, among other aspects. A general answer to these questions that allows an accurate decision implies a comprehensive interlacement of several qualitative and quantitative as well as social approaches to cooperation. This is still a necessity. The author hopes to have made a contribution to a better understanding of the benefits of cooperative planning under the perspective of the ‘era of network competition’. “The key to success in this new competitive framework, it can be argued, is the way in which this network of alliances and suppliers are welded together in partnership to achieve mutually beneficial goals.” (Christopher, 1998, p. 272).

260

6. Summary and Conclusions

“[I]t is critical for the organization […] to consider whether the time has come to reconfigure the chain to leverage the strengths of supply chain partners and alliances. One thing is for certain: companies which believe that they can continue to conduct ‘business as usual’ will find that their prospects for success in tomorrow’s marketplace will decline rapidly.” (Christopher, 1998, p. 284).

References

ADAC (2004). ADAC-AutoMarxX – Juni 2004. Retrieved February 15, 2005, from http://www.adac.de/mitgliedschaft_leistungen/motorwelt/automarxx/default.asp?Compone ntID=84637&SourcePageID=7479%230 AHD (2000). Competition. The American Heritage Dictionary of the English Language. 4th ed. Retrieved February 15, 2005, from http://www.bartleby.com/61/69/C0526900.html AHD (2000a): Cooperation. The American Heritage Dictionary of the English Language. 4th edition. Retrieved February 15, 2005, from http://www.bartleby.com/61/46 /C0624600.html Ahn, T. K., Ostrom, E., Schmidt, D., Shupp, R., & Walker, J. (2001). Cooperation in PD games: Fear, greed, and history of play. Public Choice, 106.1-2, 137-155. Ahsen, A.v., Buxmann, P., Martín Díaz, L., & Wolf, K. (2002). Supply Chain Management Software: A Survey for the SCM Software Market in the European Automotive Industry. Research Report, Freiberg University of Technology, Chair of Information Systems. AI (2002, September 20). Newswire – Mixed. Retrieved February 15, 2005, from http://www. automobilindustrie.de/news/ai_beitrag_36357.html AI (2005, February 7). Bosch liefert wieder. Retrieved February 15, 2005, from http://www. automobilindustrie.de/news/ai_beitrag_1757457.html Andel, T. (1995). There’s Power in Numbers. Transportation & Distribution, 36.8, 67-72. Andel, T. (1999). Centralization of stocks: Retailers vs. manufacturer. Management Science, 45.2, 178-192.

262

References

Anonymous (2000, August 21). Dell aims web-based SCM program at forecasts. Electronic Buyer News, 1225, 70. Anonymous (n.d.). Kosten-Nutzen-Rechnung Lieferkettenmonitoring Automobilindustrie. [internal document]. Ansari, A. & Heckel, J. (1987). JIT purchasing: Impact of freight and inventory costs. Journal of Purchasing and Materials Management, 23.2, 24-28. Anupindi, R. & Bassok, Y. (1999). Centralization of stocks: Retailers vs. manufacturer. Management Science, 45.2, 178-192. Apte, U. M. & Viswanathan, S. (2000). Effective cross docking for improving distribution efficiencies. International Journal of Logistics: Research & Applications, 3.3, 291-302. Arnold, U. (1995). Beschaffungsmanagement. 1st ed. Stuttgart: Schäffer-Poeschel. Arnold, D., Isermann, H., Kuhn, A., & Tempelmeier, H. (2004). Handbuch Logistik. 2nd ed. Berlin et al.: Springer. Arrighetti, A., Bachmann, R., & Deakin, S. (1997). Contract law, social norms and inter-firm cooperation. Cambridge Journal of Economics, 21.2, 171-195. Audi AG (2003). Audi group and locations. Retrieved November 15, 2003, from Audi Intranet. Audi AG (2003e). Vertriebsregionen. Retrieved November 15, 2003, from Audi Intranet. Audi AG (2003f). FAQ – Welche Zerlegungsgrade gibt es? Retrieved November 15, 2003, from Audi Intranet. Audi

AG

(2005).

2004

Annual

Report.

Retrieved

February

15,

2005,

from

http://www.audi.com/etc/medialib/cms4imp/audi/company/financial_information /2005.Par.0001.File.pdf Audi AG (n.d. a). Audi tradition – An overview. Retrieved November 15, 2003, from Audi Intranet. Audi AG (n.d. b). Turbulent times. Retrieved November 15, 2003, from Audi Intranet. Audi AG (n.d. c). Production location worldwide. Retrieved November 15, 2003, from Audi Intranet.

References

263

Audi AG (n.d. d). Production locations Great Britain. Retrieved November 15, 2003, from Audi Intranet. Audi AG (n.d. e). Production Plants Malaysia. Retrieved November 15, 2003, from Audi Intranet. Audi AG (n.d. f). Volkswagen Transport – Strategische Zielsetzung. Retrieved November 15, 2003, from Audi Intranet. Audi Hungaria Motor Kft. (n.d.). Standard slides for company presentation of Audi Hungaria Motor Kft.November 15, 2003Audi USA (n.d.). Audi A8 L Gallery. Retrieved February 15, 2005,

from

http://www.audiusa.com/model_gallery/0,,usageId-ExteriorLarge_modelId-

200413_ status-P_countrycode-1_mediaId-960058_,00.html AUDIWORLD (2003a). Audi A8 4.0 TDI quattro. Retrieved February 15, 2005, from http://audiworld.com/news/03/geneva/a8/content.shtml AUDIWORLD (2003b). Audi A8 4.0 TDI quattro. Retrieved February 15, 2005, from http://audiworld.com/news/03/geneva/a8/content2.shtml Autohaus-Online (2003). VW/Audi-Verträge: Entscheidung rückt näher. Retrieved February 15, 2005, from http://www.autohaus-online.de/news.php/53705,1324312352 Axelrod, R. (1997). The Complexity of Cooperation: Agent-Based Models of Competition and Collaboration. Princeton, NJ: Princeton University Press. Axelrod, R. (2000). On Six Advances in Cooperation Theory. Analyse & Kritik, 22, 130-151. Baganha, M. P., & Cohen, M. A. (1998). The stabilizing effect of inventory in supply chains. Operations Research, 46.3, S72-S83. Bagchi, P. K. & Virum, H. (1996). European logistics alliances: A management model. The International Journal of Logistics Management, 7.1, 93-108. Bagchi, P. K. & Virum, H. (1998). Logistical alliances: Trends and prospects in integrated Europe. Journal of Business Logistics, 19.1, 191-213. Balinski, M. & Quandt, R. (1964). On an integer program for a delivery problem. Operations Research, 12, 300-304.

264

References

Ballou, R. H. (1999). Business logistics management. 4th ed. Upper Saddle River, NJ: Prentice Hall. Ballou, R. H., Gilbert, S. M., & Mukherjee, A.(2002). New managerial challenges from supply chain opportunities. Industrial Marketing Management, 29.1, 7-18. Banks, J. & Carson, S. (1984). Discrete-Event System Simulation. Englewood Cliffs, NJ: Prentice-Hall. Barnard, C. I. (1968). The Functions of the Executive. Cambridge, MA.: Harvard University Press. Barua, A., Kriebel, C. H., & Mukhopadhyay, T. (1995). Information Technologies and Business Value: An Analytic and Empirical Investigation. Information Systems Research, 6.1, 3-23. Benbasat, I., Goldstein, D.K., & Mead, M. (1987). The Case Research Strategy in Studies of Information Systems. MIS Quarterly, 11.3, 369-386. Benisch, W. (1972). Kooperative Formen der Unternehmenszusammensetzung, Kooperationen. In A. Sölter & C. Zimmerer (eds.), Handbuch der Unternehmenszusammenschlüsse (pp. 149168). Munich: Verlag moderne Industrie. Benisch, W. (1973). Kooperationsfibel. 4th ed. Bergisch-Gladbach: Bundesverband der deutschen Industrie. Bensaou, M. (1997). Interorganizational cooperation: The role of information technology – An empirical comparison of U.S. and Japanese supplier relations. Information Systems Research, 8.2, 107-124. Bensen, S. M. & Farrell, J. (1994). Choosing How to Compete: Strategies and Tactics in Standardization. Journal of Economic Perspectives, 8, 117-310. Bergmann, R. & Rawlings, C. (1998). Transport management – Future directions: redefining the role of transport. In J. Gattorna (ed.), Strategic Supply Chain Alignment (pp. 369-380). Burlington, VT: Gower. Berliner

Börse

(2003).

Kapitalmarktzins.

Retrieved

February

15,

2005,

http://www.berlinerboerse.de/encyclopedia.html?ID_DEFINITION=558&debug=

from

References

265

Berry, D., Towill, D. R., & Wadsley, N. (1994). Supply chain management in the electronics products industry. International Journal of Physical Distribution & Logistics Management, 24.10, 20-32. Bertram, B. & Hofmann, J. (2003). Im Daimler-Werk Ludwigsfelde könnten ab 2006 VW Transporter vom Band rollen. Handelsblatt, 016, 16. Retrieved November 15, 2003, from Genios database. Bidlingmaier, J. , Jacobi, H., & Uherek, E. W. (1967). Absatzpolitik und Distribution. Wiesbaden: Gabler. Bhattacharya, A. K., Coleman J. L., & Brace, G. (1996). The structure of conundrum in supply chain management. The International Journal of Logistics Management, 7.1, 39-48. Bleicher, K. (1989). Kritische Aspekte des Managements zwischenbetrieblicher Kooperation. Thexis, 3, 4-8. Blohm, H. (1980). Kooperation. In E. Grochla (ed.), Handwörterbuch der Organisation (pp. 890893). 2nd ed. Stuttgart: Schäffer. Boehme,

J.

(1986).

Innovationsförderung

durch

Kooperation:

Zwischenbetriebliche

Zusammenarbeit als Instrument des Innovationsmanagements im kleinen und mittleren Unternehmen bei der Einführung der Mikroelektronik in Produkte und Verfahren. Berlin: E. Schmidt. Boettcher , E. (1974). Kooperation und Demokratie in der Wirtschaft. Tübingen: Mohr. BOLA (n.d.). The Management of Inventory: Overview. Retrieved February 15, 2005, from http://sol.brunel.ac.uk/~jarvis/bola/operations/stock/stock.html Bölzing, D. (2004). E-Logistics: Internetbasierte Logistikprozesse und –dienstleistungen für effizientes Supply Chain Management. In H. C. Pfohl (ed.), Supply Chain Management: Logistik plus? (pp. 107-117). Darmstadt: Erik Schmidt. Bosak, J. (1997). XML, Java, and the future of the web. Retrieved February 15, 2005, from http://sunsite.unc.edu/pub/sun-info/standards/xml/why/xmlapps.html Bowersox, D. J. (1990). The strategic benefits of logistics alliances. Harvard Business Review, 68.4, 36-45.

266

References

Bowersox, D. J. (1997). Integrated supply chain management: A strategic imperative. Annual Conference Proceedings (pp. 181-189). Chicago, IL: Council of Logistics Management. Bowersox, D. J. & Closs, D. (1996). Logistical management: The integrated supply chain. New York, NY: McGraw-Hill. Bowman, R. J. (2000). Ford: How to maintain a 10,000-mile supply line. Retrieved February 15, 2005, 2004, from http://www.supplychainbrain.com/archives/07.00.ford.htm? adcode=5 Bramel, J. & Simchi-Levi, D. (1997). The Logic of Logistics: Theory, Algorithms, and Applications for Logistics Management. New York, NY: Springer. Brandenburger, A. M. & Nalebuff, B. (1996). Co-opetition - A revolutionary mindset that combines competition and cooperation; The game theory strategy that's changing the game of business. New York, NY: Currency/Doubleday. Brewer, P. C. & Speh T. W. (2000). Using the balanced scorecard to measure supply chain performance. Journal of Business Logistics, 21.1, 75-95. Büchs, M. J. (1991). Zwischen Markt und Hierarchie. Zeitschrift für Betriebswirtschaft, 61, Ergänzungsheft 1, 1-38. Burt, D. (1984). Proactive Procurement. Englewood Cliffs, NJ: Prentice-Hall. Burton, T. T. (1988). JIT/repetitive sourcing strategies: Tying the knot with your suppliers. Production and Inventory Management Journal, 29.4, 38-42. Buxmann, P. (1996), Standardisierung betrieblicher Informationssysteme. Wiesbaden: Gabler. Buxmann,

P.

(2001).

Informationsmanagement

in

vernetzten

Unternehmen:

Wirtschaftlichkeitsanalyse, Organisationsänderungen, Erfolgsfaktor Zeit. Wiesbaden: Gabler. Buxmann, P. (2001a). Network Effects on Standard Software Markets. In Proceedings of the 2nd IEEE Conference on Standardization and Innovation in Information Technology (SIIT 2001). October 2001, Boulder, CO. Buxmann, P., Ahsen, A. v., Martín Díaz, L., & Wolf, K. (2004b). Usage and Evaluation of Supply Chain Management Software: Results from an Empirical Study in the European Automotive Industry. Information Systems Journal 14.3, 295-309.

References

267

Buxmann, P. & König, W. (2000). Zwischenbetriebliche Kooperationen auf Basis von SAPSystemen – Perspektiven für die Logistik und das Servicemanagement. Berlin et al.: Springer. Buxmann, P., König, W., Fricke, M., Hollich, F., Martín Díaz, L., & Weber, S. (2004). InterOrganizational Cooperation with SAP Solutions: Design and Management of Supply Networks. 2nd ed. Berlin et al.: Springer. Buxmann, P., Martín Díaz, L., & Ahsen, A. v. (2003). Kooperationen in Supply-Chains Ökonomische Ansätze und Anwendung eines Simulationsmodells. Wirtschaftsinformatik 45.5, 509-514. Buxmann, P., Martín Díaz, L., & Wüstner, E. (2002). XML-based supply chain management – As SIMPLEX as it is. Proceedings of the 35th Hawaii International Conference on System Sciences (HICSS 2002). January 2002, Hawaii, HI. Buzzel, R. D., Quelch, J. A., & Salmon, W. J. (1990). The costly bargain of trade promotion. Harvard Business Review, 68.2, 141-148. Cachon, G.P. (1999). Managing Supply Chain Demand Variability with Scheduled Ordering Policies. Management Science 45.6, 843-856. Cachon, G. P. & Fisher, M. (2000). Supply Chain Inventory Management and the Value of Shared Information. Management Science, 46.8, 1032-1048. Cameron, K. S. (1984). The Effectiveness of Ineffectiveness. Research in Organizational Behavior, 6, 235-285. Cameron, K.S. (1986). Effectiveness as Paradox: Consensus and Conflict in Conceptions of Organizational Effectiveness. Management Science 32.5, 539-553. Carrell, M. R., Jennings, D. F., & Heavrin, Ch. (1996). Fundamentals of organizational behavior. Upper Saddle River, NJ: Prentice-Hall. Carter, J. R. & Ferrin, B. G. (1995). The impact of transportation costs on supply chain management. Journal of Business Logistics, 16.1, 189-212. Cavinato, J. (1991). Identifying interfirm total cost advantages for supply chain competitiveness. International Journal of Purchasing and Materials Management, 27.4, 10-15.

268

References

Celley, A. F., Clegg, W. H., Smith, A. W., & Vonderembse, M. A. (1986). Implementation of JIT in the United States. Journal of Purchasing and Materials Management, 22.1, 9-15. Chao, I. M., Golden, B. L., & Wasil, E. (1993). A new heuristic for the multi-depot vehicle routing problem that improves upon best-known solutions. American Journal of Mathematical and Management Science, 13.3, 371-406. Chapman, S. N. & Carter, P. L. (1990). Supplier/customer inventory relationships under just-intime. Decision Sciences, 21.1, 35-51. Chen, F., Drezner, Z., Ryan, J. K., & Simchi-Levi, D. (1998). Quantifying the bullwhip effect: The impact of forecasting, lead times and information. Working Paper, Department of Industrial Engineering and Management Science, Northwestern University, Evanston, IL. and School of Industrial Engiennering, Purdue University, Lafayette, IN. Chen, F., Drezner, Z., Ryan, J. K., & Simchi-Levi, D. (1999). The Bullwhip Effect: Managerial Insights on the Impact of Forecasting and Information on Variability in a Supply Chain. In S. Tayur, R. Ganeshan, & M. Magazine (eds.), Quantitative Models for Supply Chain Management (pp. 417-439). Boston et al.: Kluwer Academic Publishers. Chen, F., Drezner, Z., Ryan, J. K., & Simchi-Levi, D. (2000). Quantifying the bullwhip effect in a simple supply chain: The impact of forecasting, lead times, and information. Management Science, 46.3, 436-443. Chen, F., Ryan, J. K., & Simchi-Levi, D. (2000a). The Impact of Exponential Smoothing Forecasts on the Bullwhip Effect. Naval Research Logistics, 47.4, 269-286. Chen, R. S., Wang, H. C., & Wang, C. M. (2003). A web-based data extraction system for supply chain management using SAP R/3. International Journal of Computer Applications in Technology, 16.1, 33-38. Christopher, M. G. (1992). Logistics and supply chain management. London: Pitman Publishing. Christopher, M. G. (1998). Relationships and alliances: Embracing the era of network competition. In J. Gattorna (ed.), Strategic supply chain management (pp. 272-284). Hampshire: Gower Press. Christopher, M. G. (1998a). Logistics and Supply Chain Management: Strategies for Reducing Cost and Improving Service. 2nd ed. London et al.: Financial Times Prentice Hall.

References

269

Christopher, M. G. & Jüttner, U. (2000). Supply Chain Relationships: Making the Transition to Closer Integration. International Journal of Logistics: Research and Applications, 3.1, 1-23. Chu, C.-H., He, X. J., & Hayya, J. C. (2002). A Data Warehousing Approach to Supply Chain Integration. September 15, 2004, from http://supplychain.ittoolbox.com/documents/ document.asp?i=1752. Chung, P. T. & Ou, J. (2001). Impact on inventory costs with consolidation of distribution centers. IIE Transactions, 33.2, 99-121. Clark, A. & Scarf, H. (1960). Optimal policies for a multi-echelon inventory problem. Management Science, 6.4, 475-490. Clarke, G. & Wright, J. W. (1964). Scheduling of vehicles from a central depot to a number of delivery points. Operations Research, 12.4, 568-581. Coase, R. H. (1937). The nature of the firm. Economica, 4.16, 386-405. Columbia Encyclopedia (2002). Kant, Immanuel. Retrieved February 15, 2005, from http://www.bartleby.com/65/ka/Kant-Imm.html Cooper, M. C. & Ellram, L. M. (1993). Characteristics of supply chain management and the implications for purchasing and logistics strategy. International Journal of Logistics Management, 4.2, 13-24. Cooper, M. C., Ellram, L. M., & Hanks, A. M. (1997a). Meshing multiple alliances. Journal of Business Logistics, 18.1, 67-89. Cooper, M. C., Lambert, D. M., & Pagh, J. D. (1997b). Supply chain management: More than a new name for logistics. The International Journal of Logistics Management, 8.1, 1-14. Cox, A. W. (1997). Business success: A Way of Thinking about Strategy, Critical Supply, Chain Assets and Operational Best Practice. Midsomer Norton, Bath: Earlsgate Press. CPFR

(n.d.).

Glossary

of

terms.

Retrieved

February

15,

2005,

from

http://www.cpfr.org/documents/pdf/CPFR_Tab_10.pdf Creswell, J. W. (1994). Research Design: Qualitative and Quantitative Approaches. Thousand Oaks, CA: Sage.

270

References

Croom, S., Romano, P., & Giannakis, M. (2000). Supply chain management: An analytical framework for critical literature review. European Journal of Purchasing & Supply Management, 6.1, 67-83. Croxton, K. L., García-Dastugue, S. J., Lambert, D. M., & Rogers, D. S. (2001). The supply chain management processes. The International Journal of Logistics Management, 12.2, 13-36. Cusumano, M. A. & Takeishi, A. (1991). Supplier relations and management: A survey of Japanese, Japanese-transplant, and U.S. auto plants. Strategic Management Journal, 12.8, 563588. Daft, R. L. (2001). Organization theory and design. 7th ed. Cincinnati, OH: South-Western Publishing. Dantzig, G. B. & Ramser, J. H. (1959). The Truck Dispatching Problem. Management Science, 6.1, 80-91. Darke, P., Shanks, G., & Broadbent, M. (1998). Successfully completing case study research: combining rigour, relevance and pragmatism. Information Systems journal, 8.4, 273-290. Darrow, B. (2002): Software spending soars among IT-Savvy SMBs. Computer Reseller News, No. 1006, 39-49. Davis, D. (1995). Third party delivers. Manufacturing Systems, 13.8, 66-68. Davis, T. (1993). Effective supply chain management. Sloan Management Review, 34.4, 35-46. Dawe, R. L. (1997). Move it fast...eliminate steps. Transportation & Distribution, 38.9, 67-74. DB Cargo (n.d.). DB Cargo Zusatz-Dienstleistung zu Logistikzügen: Logistic Center Hungaria Kft. Übernimmt Rangierdienst bei Audi Györ. Retrieved November 15, 2003, from: Audi Intranet. Deakin, E. (1988). Supplier management in a just-in-time inventory system. Journal of Accountancy, 166.6, 128-131. Deardorff, A. V. (2001). Glossary of internationals economics. Retrieved February 15, 2005, from http://www-personal.umich.edu/~alandear/glossary/ Deierlein, B. (2000). JIT: Zero tolerance for late deliveries. Fleet Equipment, 26.1, 36-40.

References

271

Domschke, W. (1997). Logistik: Rundreisen und Touren. 4th ed. Munich: Oldenbourg. Domschke, W. & Drexl, A. (1996). Logistik: Standorte. 4th ed. Munich: Oldenbourg. Domschke,

W.;

Mayer,

G.,

&

Wagner,

B.

(2002).

Effiziente

Modellierung

von

Entscheidungsproblemen: Das Beispiel des Standardisierungsproblems. Zeitschrift für Betriebswirtschaft, 72.8, 847-863. Domschke, W. & Drexl, A. (2005). Einführung in Operations Research. 6th ed. Berlin et al.: Springer. Domschke, W. & Wagner, B. (2005). Models and Methods for Standardization Problems. European Journal of Operational Research, 162.3, 713-726. Dornier, P. P., Ernst, R., Fender, M., & Jouvelis, P. (1998). Global operations and logistics – text and cases. New York, NY: John Wiley & Sons. DRA Systems (2000). Package drasys.or.graph. Retrieved February 15, 2005, from http://opsresearch.com/OR-Objects/api/drasys/or/graph/package-summary.html Drucker, P. F. (1998). Management's new paradigms. Forbes Magazine, 162.7, 152-177. Dubé, L. & Paré, G. (2003). Rigor in Information Systems Positivist Case Research: Current Practices, Trends, and Recommendations. MIS Quarterly, 27.4, 597-636. Dyer, J. H. (1996). Does governance matter? Keiretsu alliances and assets specifity as sources of Japanese competitive advantage. Organization Science, 7.6, 649-666. Dyer, J. H., Cho, D., & Chu, W. (1998). Strategic supplier segmentation: The next “best practice” in supply chain management. California Management Review, 40.2, 57-77. Ellram, L. M. (1990). The supplier selection decision in strategic partnerships. Journal of Purchasing and Materials Management, 25.3, 8-14. Ellram, L. M. (1991). Supply chain management: The industrial organization perspective. International Journal of Physical Distribution and Logistics Management. 21.1, 13-22. Ellram, L. M. (1992). International purchasing alliances: An empirical study. The International Journal of Logistics Management, 3.1, 23-36.

272

References

Ellram, L. M. & Cooper, M. (1990). Supply chain management, partnerships, and the shipper third party relationship. The International Journal of Logistics Management, 2.2, 1-10. Ellram, L. M. & Cooper, M. (1993). The Relationship between Supply Chain Management and Keiretsu. The International Journal of Logistics Management, 4.1, 1-12. Emmerich, V. (2001). Kartellrecht. 9th ed. Munich: C. H. Beck. Encyclopaedia Britannica Online (2003). Effective. Retrieved February 15, 2005, from https://80-search.eb.com.hokhmah.s tmarys-ca.edu:2048/dictionary Endress, R. (1991). Strategie und Taktik der Kooperation: Grundlage der zwischen- und innerbetrieblichen Zusammenarbeit. Berlin: Schmidt. Eschenburg, R. (1971). Ökonomische Theorie der genossenschaftlichen Zusammenarbeit. Tübingen: Mohr. Evangelista, P. & Morvillo, A. (1999). Alliances in liner shipping: An instrument to gain operational efficiency or supply chain integration?. International Journal of Logistics: Research and Applications, 2.1, 21-38. Evans, G. N., Naim, M. M., & Towill, D. R. (1993). Dynamic supply chain performance: Assessing the impact of information systems. Logistics Information Management, 6.4, 15-25. Evers, P. T. & Beier F. J. (1998). Operational aspects of inventory consolidation decision making. Journal of Business Logistics, 19.1, 173-189. Farmer, D. & van Amstel, M. J. P. (1990). Controlling the logistics pipeline. Internationals Journal of Logistics Management, 1.1, 19-27. Farrell, J. & Saloner, G. (1985). Standardization, Compatibility, and Innovation. Rand Journal of Economics, 16.1, 70-83. Fawcett, S. E. & Magnan, G. M. (2001). Achieving world-class supply chain alignment: Benefits, barriers,

and

bridges.

Retrieved

February

15,

2005,

from

http://www.

capsresearch.org/ReportPDFs/AchievingSCAlignment.pdf Fawcett, S. E. & Magnan, G. M. (2002). The rhetoric and reality of supply chain integration. International Journal of Physical Distribution & Logistics Management. 32.5, 339-361.

References

273

Ferrin, B. (1994). Planning just-in-time supply operations: A multiple-case analysis. Journal of Business Logistics, 15.1, 53-70. Fisher, M. L. (1997). What is the right supply chain for your product?. Harvard Business Review, 75.2, 105-116. Fleischmann, B. (2002) Begriffliche Grundlagen. In D. Arnold, H. Isermann, A. Kuhn, & H. Tempelmeier (eds.), Logistik Handbuch (pp. A1-3-A1-13). Berlin et al.: Springer. Forrester, J. (1961). Industrial Dynamics. New York, NY: John Wiley & Sons. Foster, T. A. (1993). Brewing up a logistics partnership. Distribution, 92.9, 49-52. Frankel, R. & Whipple, J. S. (1996). Alliance formation motives: A comparison of international perspectives. The International Journal of Logistics Management, 7.2, 19-30. Fransoo, J. C. & Wouters, M. J. F. (2000). Measuring the bullwhip effect in the supply chain. Supply Chain Management, 5.2, 78-89. Frese, E. (2000). Grundlagen der Organisation: Konzept - Prinzipien - Strukturen. 8th ed. Wiesbaden: Gabler. Friedman, M. (2002). Working out supply chain links. Computing Canada, 28.23, 22-24. Ganeshan, R., Jack, E., Magazine, M. J., & Stephens, P. (1999). A taxonomic review of supply chain management research. In S. Tayur, R. Ganeshan, & M. Magazine (eds.) Quantitative Models for Supply Chain Management (pp. 840-879). Boston et al.: Kluwer Academic Publishers. Gavirneni, S., Kapuscinski, R. & Tayur, S. (1999). Value of Information in Capacitated Supply Chains. Management Science, 45.1, 16-24. Gebauer, J., & Buxmann, P. (2000). Assessing the value of interorganizational systems to support business transactions. International Journal of Electronic Commerce, 4.4, 61-82. Gentry, J. J. (1996). The role of carriers in buyer-supplier strategic partnerships: A supply chain management approach. Journal of Business Logistics, 17.2, 33-55. Gerth, E. (1971). Zwischenbetriebliche Kooperation. Stuttgart: Poeschel.

274

References

Giannoccaro, I. & Pontrandolfo, P. (2001). Models for supply chain management. In Proceedings of the Twelfth Annual Conference of the Production and Operations Management Society (POM-2001). April 2001, Orlando, FL. Gillet, B. E. & Miller, L. R. (1974). A heuristic algorithm for the vehicle-dispatch problem. Operations Research, 22.2, 340-349. Giunipero, L. C. & Brand, R. C. (1996). Purchasing's role in supply chain management. International Journal of Logistics Management, 7.1, 29-38. Goldberg, V. P. (1976). Regulation and administered contracts. Bell Journal of Economics, 7.2, 426-448. Golden, B. L., Magnanti, T. L., & Nguyen, H. Q. (1977). Implementing Vehicle Routing Algorithms. Networks, 7, 149-183. Gomes, R. & Mentzer, J. T. (1988). A systems approach to the investigation of just-in-time, Journal of Business Logistics, 9.2, 71-88. Goodman, L. A. (1961). Snowball sampling. Annals of Mathematical Statistics, 32.1, 148-170. Gowa, J. (1986). Anarchy, Egoism, and Third Images: The Evolution of Cooperation and International Relations. International Organization, 40.1, 167-186. Grant, R. M. (1998). Contemporary Strategy Analysis. 3rd ed. Malden, MA: Blackwell. Grochla, E. (1969). Betriebsverbindungen. Berlin: de Gruyter. Grossmann, S. & Hart, O. (1983). An analysis of the principal-agent problem. Econometrica, 51.1, 7-45. Grunwald,

W.

(1982).

Konflikt-Konkurrenz-Kooperation:

Eine

theoretisch-empirische

Konzeptanalyse. In W. Grunwald & H. G. Lilge (eds.), Kooperation und Konkurrenz in Organisationen (pp. 50-96). Bern et al.: Haupt. Guetzkow, R H. (1966). Relations among Organizations. In R. Bowers (ed.), Studies In Behavior In Organizations. Athens, GA: University of Georgia Press. Hahn, C. K., Pinto, P. A., & Bragg, D. J. (1983). “Just-in-time” production and purchasing. Journal of Purchasing and Materials Management, 19.3, 2-10.

References

275

Hall, C. (2002). Vendor Managed Inventory – Promising Value for the Trucks Parts Industry. Retrieved February 15, 2005, 2004, from: http://www.logisticcases.com/dc_cases/ value_truck_parts.pdf Hamel, G., Doz, Y. L., & Prahalad, C. K. (1989). Collaborate with Your Competitors – and Win. Harvard Business Review, 67.1, 133-139. Hansen, U., Raffée, H., Riemer, M., & Segler, K. (1983). Kooperation zwischen japanischen und Deutschen Unternehmen. Arbeitspapier 23, Institut für Marketing, Mannheim University. Harland, C. M. (1996). Supply chain management: Relationships, chains and networks. British Journal Of Management, 7 (Special Issue), 63-80. Harms, V. (1973). Interessenlage und Interessenkonflikte bei der zwischenbetrieblichen Kooperation. Würzburg et al.: Physika-Verlag. Hartley, J. J. (2000). Collaborative value analysis: Experiences from the automotive industry. Journal of Supply Chain Management, 36.4, 27-32. Hauser, H. (1981). Joint Ventures: Sonderlösungen für einzelfälle oder allgemein verwendbare Instrumente der internationalen Kooperation? Außenwirtschaft, 36, 176-194. Hax, A. C. & Candea, D. (1984). Production and Inventory Management. Englewood Cliffs, NJ: Prentice-Hall. Heide, J. B. & John, G. (1990). Alliances in industrial purchasing: The determinants of joint action in buyer-supplier relationships. Journal of Marketing, 27.1, 24-36. Heizer, J. & Render, B. (1999). Principles of Operations Management. 5th ed. Upper Saddle River, N.J: Prentice-Hall. Hellingrath, B. (2001). Business Case Lieferkettenmonitoring. [internal document].Helms, M. M., Ettkin, L., & Chapman, S. (2000). Supply chain forecasting. Collaborative forecasting supports supply chain management. Business Process Management, 6.5, 392-407. Helper, S. (1991). How much has really changed between U.S. automakers and their suppliers?. Sloan Management Review, 32.4, 15-28.

276

References

Hendrick, T. E. (1997). Purchasing consortiums: Horizontal alliances among firms buying common goods and services what? who? why? how?. Retrieved February 15, 2005, from http://www.capsresearch.org/ReportPDFs/ConsortAll.pdf Henkoff, R. (1994). Delivering the goods. Fortune, 130.11, 64-78. Henzel, F. (1968). Kooperation von Produktionsbetrieben. Zeitschrift für Betriebswirtschaft, 38, 795-804. Heskett, J. L. (1977). Logistics - essential to strategy. Harvard Business Review, 55.6, 85-96. Hines, P. (1994). Internationalization and localization of the kyoryoku kai: The spread of best practice supplier development. The International Journal of Logistics Management, 5.1, 67-72. Hines, P. (1995). Network sourcing: A hybrid approach. International Journal of Purchasing and Materials Management, 31.2, 17-24. Hines, P., Jones, D., & Lamming, R. (2000). Value stream management: Strategy and excellence in the supply chain. Upper Saddle River, NJ: Prentice-Hall. Hinsenkamp, O. (2002a, October 31). Abkürzungs- und Terminologieverzeichnis des Geschäftsbereiches Produktion. Retrieved November 15, 2003, from Audi Intranet. Hinsenkamp, O. (2002b, October 31). Abkürzungs- und Terminologieverzeichnis des Geschäftsbereiches Produktion. Retrieved November 15, 2003, from Audi Intranet. Hise, R. T. (1995). The implications of time-based competition on international logistics strategies. Business Horizons, 38.5, 39-45. Ho, S. K. M. (1999). Operations and Quality Management. London: International Thomson Business Press. Hoffmann, R. (1996). The ecology of cooperation – A systematic computer simulation approach to the factors that affect the emergence of cooperative behavior in populations of finite automata playing the repeated prisoner’s dilemma (Dissertation). Retrieved February 15, 2005, from: http://www.nottingham.ac.uk/~lizrh2/hoffmann.html Hoffmann, R. (2001). The ecology of cooperation. Theory and Decision, 50.2, 101-118.

References

277

Hofmann, J. (2002, September 11). Daimler-Chrysler und VW verlängern Zusammenarbeit. Handelsblatt, 175, 13. Holmes, G. & Darivemula, S. (2004): Driving Growth Through Collaboration. Automotive News Europe, 9.13, 2-8. Holmstrom, B. (1979). Moral hazard and observability. Bell Journal of Economics, 10.1, 74-91. Holte,

K.

(2001):

Integrierte

Planung

logistischer

Netzwerke:

Methoden

und

Modellierungsansätze zur Entscheidungsunterstützung (Dissertation). St. Gallen University. Houlihan, J. (1985). International supply chain management. International Journal of Physical Distribution and Logistics Management, 15.1, 22–38. Houlihan, J. (1987). International supply chain management. International Journal of Physical Distribution and Logistics Management, 17.2, 51–66. Hueck. G. (2003). Gesellschaftsrecht: Ein Studienbuch. 20th ed. Munich: Beck. ICON (2003a). ICON – Der Spezialist für Supply Chain Management. Retrieved November 15, 2003, from http://www.icon-germany.de ICON (2003b). SCC 4 Online Help. Retrieved November 15, 2003, from http://scc.icongermany.de/sccdemo40/JSP/LoginPage/LoginCheck.jsp ICON (2003c). Unsere Kunden. Retrieved November 15, 2003, from http://www.icongermany.de ICON (2003d). GEKO 2003. Retrieved November 15, 2003, from http://scc.icongermany.de/audi Jarillo, J.C. (1988). On strategic networks. Strategic Management Journal, 9.1, 31-41. Jarillo, J. C. & Stevenson, H. H. (1991). Co-operative strategies: The payoffs and the pitfalls. Long Range Planning, 24.1, 64-70. Jervis, R. (1988). Realism, Game Theory, and Cooperation. World Politics 40.3, 317-349. Johnson, J. C., Wood, D. F., Wardlow, D. L., & Murphy Jr., P. R. (1999) Contemporary logistics. 7th ed. Upper Saddle River, NJ: Prentice Hall.

278

References

Johnson, P. L. (2000) Transaction costs. A glossary of political economy terms. Retrieved February 15, 2005, from: http://www.auburn.edu/~johnspm/gloss/index.html?http:// www.auburn.edu/~johnspm/gloss/transaction_costs.html Johnston, H. R. & Lawrence, P. (1988). Beyond vertical integration – The rise of the value-adding partnership. Harvard Business Review, 66.4, 94-101. Jones, T. & Riley, D. (1985). Using inventory for competitive advantage through supply chain management. International Journal of Physical Distribution and Logistics Management, 15.5, 16-22. Joshi, K. & Campell, J. F. (1991). Managing inventories in a JIT environment. International Journal of Purchasing and Materials Management, 27.2, 32-37. Jünemann, R. (1989). Materialfluss und Logistik – Systemtechnische Grundlagen mit Praxisbeispielen. Berlin, Heidelberg: Springer Verlag. Kalton, G. (1983). Introduction to Survey Sampling: Quantitative Applications in the Social Sciences. Newbury Park, CA: Sage. Katz, D. & Kahn, R. L. (1978). The social psychology of organizing. 2nd ed. New York, NY: John Wiley & Sons. Katz, M. L.; Shapiro, C. (1985): Network externalities, competition, and compatibility. The American Economic Review, 75.3, 424-440. Kenneth, O. (1986). Explaining Cooperation under Anarchy: Hypotheses and Strategies. In O. Kenneth (ed.) Cooperation under Anarchy (pp. 1-24), Princeton, NJ: Princeton University Press. Kerstetter, J. (2001): Software Highfliers. Business Week, 81.26, 109-110. Kieser, A. & Walgenbach, P. (2003). Organisation. 4th ed. Stuttgart: Schäffer-Poeschel. Kimms, A. (2003). Costing Communication Standards in Information Systems Using a Minimum Cut Approach. Journal of the Operational Research Society, 54.4, 426-431. Kinkel, S. & Lay, G. (2003). Fertigungstiefe – Ballast oder Kapital? Stand und Effekte von Outund Insourcing im Verarbeitenden Gewerbe Deutschlands. Mitteilungen aus der

References

Produktionsinnovationserhebung,

279

30.

Retrieved

February

15,

2005,

from

http://www.isi.fhg.de /pi/dokumente/pi30.pdf Klein, R. (1989). Concrete and abstract Voronoi Diagrams. Berlin et al.: Springer. Klein, B., Crawford, R. G., & Alchian, A. A. (1978). Vertical integration, appropriable rents, and the competitive contracting process. Journal of Law and Economics, 21.2, 297-326. Kluck, D. (1998). Materialwirtschaft und Logistik: Lehrbuch mit Beispielen und Kontrollfragen. Stuttgart: Schäffer-Poeschel Verlag. Knoblich, H. (1969). Zwischenbetriebliche Kooperation: Wesen, Formen und Ziele. Zeitschrift für Betriebswirtschaft, 39.8, 497-514. Knolmayer, G., Mertens, P., & Zeier, A. (2002). Supply Chain Management based on SAP Systems, Order Management in Manufacturing Companies. Berlin et al.: Springer. Knoppe, M. (1997). Strategische Allianzen in der Kreditwirtschaft, Oldenbourg, Munich. König, W. (1994). Profil der Wirtschaftsinformatik. Wirtschaftsinformatik, 36.1, 80-81. Koch, Ch. (n.d.). The ABCs of supply chain management. Retrieved February 15, 2005, from http://www.cio.com/research/scm/edit/012202_scm.html Kogut, B. (1989). The Stability of Joint Ventures: reciprocity and Competitive Rivalry. Journal of Industrial Economics, 38.2, 183-198. Koller, M. (2000). When paranoia's a healthy sign - Sharing data over supply chain links necessitates a strong dose of caution. InternetWeek, 835, 135-136. Kopczak, L. R. (1997). Logistics partnerships and supply chain restructuring: Survey results from the U.S. computer industry. Production and Operations Management, 6.3, 226-247. Kopczak, L. R. (1999). Trends in third party logistics. Retrieved February 15, 2005, from http://www.ascet.com/documents.asp?d_ID=230# Koschnik, W. J. (1995). Encyclopedic Dictionary of Marketing. Vol. 2/Part 1. Munich et al.: Saur. Kotzab, H. & Otto, A. (n.d.). Transferring end-user orientation to physical distribution action Considering supply chain management as a logistical marketing approach. Retrieved February 15, 2005, 2004, from http://www.dcpress.com/jmb/kotzab.htm

280

References

Kuhn,

St.

(2001).

Prisoner’s

Dilemma.

Retrieved

February

15,

2005,

from:

http://plato.stanford.edu/entries/prisoner-dilemma/ La Londe, B. J. & Masters, J. M. (1994). Emerging logistics strategies: Blueprints for the next century. International Journal of Physical Distribution and Logistics Management, 24.7, 35-47. Lake, D. & Powell R. (1999). Strategic Choice and International Relations. Princeton, NJ: Princeton University Press. Lambert, D. M. (2002). Supply chain management. The Global Supply Chain Forum. Retrieved February

15,

2005,

from

http://fisher.osu.edu/logistics/researchpublications/

supply_chain_management/SupplyChainManagement.doc Lambert, D. M., Cooper, M. C., & Pagh, J. D. (1998). Supply chain management: Implementation issues and research opportunities. The International Journal of Logistics Management, 9.2, 119. Lambert, D. M., Emmelhainz, M. A., & Gardner, J. T. (1999). Building successful logistics partnerships. Journal of Business Logistics, 20.1, 165-181. Lamming, R. C. (1993). Beyond partnership: Strategies for innovation and lean supply. Hemel Hempstead: Prentice-Hall. Landeros, R. & Monczka, R. M. (1990). Cooperative buyer-seller relationships and a firm's competitive posture. Journal of Purchasing and Materials Management, 25.2, 8-14. Landry, S., Trudel, Y., & Diorio, M. O. (1998). Just-in-time supply: cooperation, competition, and abuse. Competitiveness Review, 8.1, 37-46. Langley, J. Jr. & Holcomb, M. C. (1992). Creating logistics customer value. Journal of Business Logistics, 13.2, 1-28. Lapide, L. & Davis, W. (2003). The Supply Chain Management Applications Report, 2002-2007. AMR

Research

Report.

Retrieved

February

15,

2005,

from

http://www.amrresearch.com/Content/view.asp?pmillid=16016&bhsh=1024&bhsw=1280& bhqs=1 Laporte, G., Nobert, Y., & Taillefert, S. (1988). Solving a family of multi-depot vehicle routing and location-routing problems. Transportation Science, 22.3, 161-172.

References

281

Law, A. M. & Kelton, W. D. (1991). Simulation Modeling & Analysis. 2nd ed. New York, NY, et al.: McGraw-Hill. Law, A. M. & McComas, M. G. (1990). Secrets of Successful Simulation Studies. Industrial Engineering, 22, 47-53. Lee, A. S. (1989). A Scientific Methodology for MIS Case Studies. MIS Quarterly, 13.1, 33-52. Lee, H. L & Billington, C. (1992). Managing supply chain inventory: Pitfalls and opportunities. Sloan Management Review, 33.3, 65-73. Lee, H. L., Padmanabhan, V., & Whang, S. (1997a). Information distortion in a supply chain: The bullwhip effect. Management Science, 43.4, 546-558. Lee, H. L., Padmanabhan, V., & Whang, S. (1997b). The bullwhip effect in supply chains. Sloan Management Review, 38.3, 93-102. Lee, H. L., So, K. C., & Tang, C. S. (2000). The Value of Information Sharing in a Two-Level Supply Chain. Management Science, 46.5, 626-643. Lewis, J. D. (1995). The connected corporation. New York, NY: Free Press. Lieb, R. C. (1992). The use of third-party logistics services by large American manufacturers. Journal of Business Logistics, 13.2, 29-42. Linné, H. (1993). Wahl geeigneter Kooperationspartner. Frankfurt am Main: Peter Lang. Littig, P. (1999). Coopetition, Die Klugen vergrößern den Kuchen. Bielefeld: Bertelsmann. Little, J. D. C., Murty, K. G., Sweeney, D. W., & Karel, C. (1963). An algorithm for the traveling salesman problem. Operations Research, 11.6, 972-989. Luderer,

B.,

Nollau,

K.,

&

Vetters,

K.

(1999).

Mathematische

Formeln

für

Wirtschaftswissenschaftler. Stuttgart et al.: B.G. Taubner. Mabert, V. A. & Venkataramanan, M. A. (1998). Special research focus on supply chain linkages: Challenges for design and management in the 21st century. Decision Sciences Journal, 29.3, 537-552. Macbeth, D. & Ferguson, N. (1994). Partnership Sourcing: An Integrated Supply Chain Approach. London: Pitman Publishing.

282

References

Macbeth, D. K., Ferguson, N., & Neil, G. (1992). Developing customer-supplier relationships. First PSERG Conference. Glasgow, England. Mally,

J.

(2002).

Lean

production.

Retrieved

February

15,

2005,

from

http://searchcio.techtarget.com/sDefinition/0,,sid19_gci810519,00.html Manager-Magazine (2003). Mercedes macht unzufrieden. Manager-Magazine.de, November 27th 2003.

Retrieved

February

15,

2005,

from

http://www.manager-

magazin.de/life/auto/0,2828,275812,00.html Manager-Magazine (2004). Toyota toppt Porsche. Manager-Magazine.de, May 27th 2004. Retrieved

February

15,

2005,

http://www.manager-magazin.de/life/auto

/0,2828,301746,00.html March, J. G., & Simon, H. A. (1993). Organizations. Cambridge, MA.: Blackwell. Markland, R. E., Vickery, S. K., & Davis, R. A. (1998). Operations management. Concepts in manufacturing and services. 2nd ed. Cincinnati, OH: South-Western College Publishing. Marschak, J. (1954). Towards an Economic Theory of Organization and Information. In R. M. Thrall, C. H. Coombs, & R. L. Davis (eds.), Decision processes (pp. 187-220). New York, NY: Wiley. Marschak, J. (1955). Elements of a Theory of Teams. Management Science, 1.2, 127-137. Martin, R. (2002). Data sharing between supply chain members. Dissertation Abstracts International, 63.11, 4009A. Marwell, G. & Schmitt, D. R. (1975). Cooperation: An Experimental Analysis. New York, NY: Academic Press. Mason-Jones, R. & Towill, D. R. (1997). Information enrichment: Designing the supply chain for competitive advantage. Supply Chain Management, 2.4, 137-148. Mathiesson-Öjmertz, B. & Johansson, M. I. (2000). Influences of Process Location on Materials Handling: Cases from the Automotive Industry. International Journal of Logistics: Research and Applications, 3.1, 25-39. McIvor, R. T., Humphreys, P. K., & McAleer, W.E. (1998). European car makers and their suppliers: changes at the interface. European Business Review, 98.2, 87-99.

References

McLaughlin,

283

B.

(n.d.):

All

About

JAXP.

Retrieved

February

15,

2005,

from

ftp://www6.software.ibm.com/software/developer /library/x-jaxp.pdf Meimberg, P. (1970). Betriebswirtschaftliche Aspekte der Kooperation. In Präsidium des Deutschen Raiffeisenverbandes (ed.), Freiheit und Ordnung (pp. 245-254). Bonn: Raffeisendruckerei. Mentzer, J. T., DeWitt, W., Keebler, J. S., Min, S., Nix, N. W., Smith, C. D., & Zacharia, Z. G. (2001). Defining supply chain management. Journal of Business Logistics, 22.2, 1-25. Metters, R. (1997). Quantifying the bullwhip effect in supply chains. Journal of Operations Management, 15.2, 80-100. Metz, P. (1997). Demystifying supply chain management: Accomplishments and challenges. In Annual Conference Proceedings (pp. 237-255). Chicago, IL: Council of Logistics Management. Mill, J. S. (1924). Grundsätze der politischen Ökonomie mit einigen ihrer Anwendungen auf die Sozialphilosophie. Jena: Fischer. Milner, H. (1992). International Theories of Cooperation among Nations: Strengths and Weaknesses. World Politics 44.3, 466-496. Mirrlees, J. (1974). Notes on welfare economics, information, and uncertainty. In M. Balch, D. Mc-Fadden, & S. Wu. (eds.), Essays on Economic Behavior under Uncertainty (pp. 243-257). Amsterdam: North-Holland. Mirrlees, J. (1976). The optimal structure of authority and incentives within an organization. Bell Journal of Economics, 7.1, 105-131. Mitchell, M. A., LeMay, S. A., Arnold, D. R., & Turner, G. B. (1992). Symbiotic logistics. The International Journal of Logistics Management, 3.2, 19-29. Monahan, J. P. (1984). A quantity discount pricing model to increase vendor profits. Management Science, 30.6, 720-726. Monoochehri, G. H. (1984). Suppliers and the just-in-time concept. Journal of Purchasing and Materials Management, 20.1, 16-21.

284

References

Monczka, R., Trent, R., & Handfield, R. (2001). Purchasing and Supply Chain Management. 2nd ed. Cincinnati, OH: South-Western College Publishing. Morgan, B. J. T. (1984). Elements of Simulation. London: Chapman and Hall. Morton, T. E. (1999). Production Operations Management. Cincinnati, OH: South-Western College Publishing. Mukhopadhyay, T. & Kekre, S. (2002). Strategic and Operational Benefits of Electronic Integration in B2B Procurement Processes. Management Science, 48.10, 1301-1313. Murphy, T. (2001). GM-Fiat gains partners. Ward's Auto World, 37.5, 53-54. Myerson, R. B. (2001). Game theory – Analysis of conflict. 4th printing. Cambridge, MS: Harvard University Press. MWD (2004): Cooperation. Merriam-Webster Online Dictionary. Retrieved February 15, 2005, from

http://www.m-w.com/cgi-bin/dictionary?book=Dictionary&va=

cooperation&x=0&y=0 N-TV (2003, November 27): Mercedes fällt in Ungnade – Unzufrieden mit deutschen Autos. Retrieved February 15, 2005, from http://www.n-tv.de/5196474.html Nassimbeni, G. (1998). Network structures and co-ordination mechanisms: A taxonomy. International Journal of Operations and Production Management, 18.6, 538-554. New, S. J. & Ramsay, J. (1995). Supply chains – corporate path to economic disaster? Fourth International IPSERA Conference, Birmingham, England. Nishiguchi, T. (1994). Strategic industrial sourcing: The Japanese advantage. Oxford: Oxford University Press. Odette (2003). Supply chain monitoring version 1.0. [internal document] Oliver, R. W. (1999). The end of inventory?. Journal of Business Strategy, 20.1, 8-11. Oliver, R. K. & Webber, M. (1982). Supply-chain management: logistics catches up with strategy. Outlook, Booz, Allen and Hamilton Inc. Reprinted in M. Christopher (ed.), Logistics: The strategic issues (pp. 63-75). 1992. London: Chapman-Hall.

References

285

OnePL (2002). Cross-dock and merge in transit. OnePL – Logistics Intelligence. Retrieved February 15, 2005, from http://www.onepl.com/onepl3_2.html Orlikowski, W. J. & Baroudi, J. J. (1991). Studying information technology in organisations: Research approaches and assumptions. Information Systems Research, 2.1, 1-28. Oxford English Dictionary (n.d.). Profitability. Retrieved February 15, 2005, from http://80dictionary.oed.com.hokhmah.stmarys-ca.edu:2048/cgi/entry/00189502 Oye, K. (1986). Cooperation under Anarchy. Princeton, NJ: Princeton University Press. Parsons, T. (2001). The Social System. New ed. Repr. London et al.: Routledge. Patton, M. Q. (1984). Data Collection: options, strategies and cautions. In L. Rutman (ed.), Evaluation Research Methods: A Basic Guide (39-63). 2nd ed. Beverly Hills, CA: Sage. Patton, M. Q. (1987). How to Use Qualitative Methods in Evaluation. Newbury Park, CA: Sage. Paul, L. G. (2003). Suspicious minds; Collaboration among trading partners can unlock great value. Mistrust is the barrier. Here are six ways to build confidence. CIO, 16.7, 74-82. Perrow, Ch. (1961). The analysis of goals in complex organizations. American Sociological Review, 26.6, 854-866. Pfohl, H. D. (2004). Grundlagen der Kooperation in logistischen Netzwerken. In H. D. Pfohl (ed.), Erfolgsfaktor Kooperation in der Logistik (pp. 1-36). Berlin: Erik Schmidt. Pfohl, H. D., Hofmann, E., & Trumpfheller, M. (2004). Zukunft des Netzwerkmanagements. Industrie Management, 20.4, 35-38. Pidd, M. (1992). Computer Simulation in Management Science. 3rd ed. New York, NY, et al.: John Wiley & Sons. Plaßmann, M. (1974). Die Kooperationsentscheidung des Unternehmers (Dissertation). Münster University. Porter, M. E. (1985). Competitive advantage: Creating and sustaining superior performance. New York, NY: Free Press. Porter, M. E. (1986). Wettbewerbsvorteile. 6th ed. Frankfurt am Main: Campus-Verlag.

286

References

Porter, M. E. (1986a). Changing Patterns of International Competition. California Management Review, 28.2, 9-40. Porter, M. E. (1989). Der Wettbewerb in globalen Märkten: Ein Rahmenkonzept. In M. E. Porter (ed.), Globaler Wettbewerb: Strategien der neuen Internationalisierung (pp. 17-68). Wiesbaden: Gabler. Porter, M. E. & Fuller, M. B. (1986). Coalitions and Global Strategy. In M. Porter (ed.), Competition in Global Industries (pp. 315-344). Boston, MS: Harvard Business School. Prahalad, C. K. & Hamel, G. (1990). The core competence of the corporation. Harvard Business Review, 68.3, 79-91. Premkumar, P. . (2000). Interorganization Systems And Supply Chain Management: An Information Processing Perspective. Information Systems Management, 17.3, 56-69. Pucik, V. (1988). Strategic Alliances, Organizational Learning, and Competitive Advantage: The HRM Agenda. Human Resource Management, 27.1, 77-93. Pugh, D. S. (1981). The Aston Program Perspective. In A. H. Van de Ven & W. F. Joyce (eds.), Perspectives on Organization Design and Behavior (pp. 135-166). New York, NY: John Wiley & Sons. Pugh, D. S. & Hickson, D. J. (1971). Eine dimensionale Analyse bürokratischer Strukturen. In R. Mayntz (ed.), Bürokratische Organisation (pp. 82-93). Cologne: Kiepenheuer & Witsch. Quinn, F. J. (2000). The Master of Design: An Interview with David Simchi-Levi. Supply Chain Management Review, November/December 2000, 75-80. Raghunathan, S. (2001). Information Sharing in a Supply Chain: A Note on its Value when Demand is Nonstationary. Management Science, 47.4, 605-610. Raia, E. (1988). JIT in Detroit. Purchasing, 105.3, 68-77. Ramsay, J. (1996). The case against purchasing partnerships. International Journal of Purchasing and Materials Management, 32.4, 13-20. Rath, H. (1990). Neue Formen der internationalen Unternehmenskooperation, Hamburg: Steuerund Wirtschaftsverlag.

References

287

Revelle, C. S. & Laporte, G. (1996). The plant location problem: New models and research prospects. Operations Research, 44.6, 864-874. Richardson, H. L. (1990). Explore outsourcing. Transportation and Distribution, 31.7, 17-19. Richardson, H. L. (1998). Pooling with competitors. Transportation and Distribution, 39.11, 105110. Risse, J., Stommel, H., & Zadek, H. (2002). Lieferkettenkennzahlen als Instrument für ein effizientes Supply Chain Controlling. In R. Hossner (ed.), Jahrbuch der Logistik 2002 (pp. 189-193). Düsseldorf: Verlagsgruppe Handelsblatt. Robbins, S. P. (1990). Organization Theory – Structure, Design, and Applications. 3rd ed. Englewood Cliffs, NJ: Prentice Hall. Rojek, D. (2004). Treffsichere Prognosen – Logistikpotentiale ausschöpfen durch Bündelung der vorhandenen Informationen. In H. C. Pfohl (ed.), Supply Chain Management: Logistik plus? (pp. 185-204). Darmstadt: Erik Schmidt. Ross, S. (1973). The economic theory of agency: The principal's problem. American Economic Review, 63.2, 134-139. Rotering, J. (1993). Zwischenbetriebliche Kooperation als alternative Organisationsform. Stuttgart: Schäffer-Poeschel. Ryan, J. K. (1997). Analysis of inventory models with limited demand information. Dissertation Abstracts International, 58.1, 6161B. Sadrian, A. A. & Yoon, Y. S. (1992). Business volume discount: A new perspective on discount pricing strategy. International Journal of Purchasing and Materials Management, 28.2, 43-46. Sahin, F. & Robinson, E. P. (2002). Flow coordination and information sharing in supply chains: Review, implications, and directions for future research. Decision Sciences, 33.4, 505-536. Sahlins, M. D. (1973). On the Sociology of Primitive Exchange. In M. Banton (ed.), The Relevance of Models for Social Anthropology (139-235). 3rd print. London: Tavistock. Salje, P. (1981). Die mittelständische Kooperation zwischen Wettbewerbspolitik und Kartellrecht. Tübingen: Mohr.

288

References

Saunders, M. J. (1995). Chains, pipelines, networks and value stream: The role, nature and value of such metaphors in forming perceptions of the task of purchasing and supply management. First Worldwide Research Symposium on Purchasing and Supply Chain Management (pp. 473-485), Tempe, AZ. Schaffer, B. (1998). Cross docking can increase efficiency. Automatic I.D., 14.8, 34-37. Schaude, G. (1991). Kooperation, Joint Venture, Strategische Allianzen. 2nd ed. Eschborn: Rationalisierungs-Kuratorium der Deutschen Wirtschaft. Scheidler, R. (2002). e-SCM Projekt “Lieferkettenmonitoring”. [internal document]. Scheidler, R. (2003a). Auswahlkriterien kritische Pfade. [internal document]. Scheidler, R. (2003b). GEKO 2003 – Ganzheitliche ertragsorientierte Kettenoptimierung. [internal document]. Schelling, T. (1963). The Strategy of Conflict. New York, NY: Oxford University Press. Schlittgen, R. & Streitberg, B. H. J. (2001). Zeitreihenanalyse. Munich et al.: Oldenbourg. Schmidt, C. (2002). K-DOB IS Produktherstellung – Prozess- und Systemdokumentation. Retrieved November 15, 2003, from Audi Intranet. Schmidt-Mohr, U. (1997). Agency-Theorie. Gabler Wirtschaftslexikon [CD-ROM]. Wiesbaden: Gabler. Schneeweiss, Ch. (1992). Planung 2: Konzepte der Prozeß- und Modellgestaltung. Berlin, Springer. Schneeweiss, Ch. (1994). Elemente einer Theorie Hierarchischer Planung. OR Spektrum, 16.2, 161-168. Schneeweiss, Ch. (1999). Hierarchies in Distributed Decision Making - An Outline of a General Theory. LS für Unternehmensplanung, insb. OR, Diskussionsarbeit, Nr. 62, 1999. Schneider, L. M. (1985). New era in transportation strategy. Harvard Business Review, 63.2, 118126. Schwartz, E. (2001). Skipping steps. InfoWorld, 23.6, 28-29.

References

289

Schwarz, P. (1979). Morphologie von Kooperationen und Verbänden. Tübingen: Mohr. Sela, A. & J. Vleugels (1997). Game theory. Retrieved February 15, 2005, from http://www.sfb504.uni-mannheim.de/glossary.game.htm Seidl,

K.

(2004).

Supply

Chain

Management

Software:

Einsatzmöglichkeiten

und

Nutzenerwartungen. In H. C. Pfohl (ed.), Supply Chain Management: Logistik plus? (pp. 161183). Darmstadt: Erik Schmidt. Semeijn, J. & Vallenga, D. B. (1995). International logistics and one-stop shopping. International Journal of Physical Distribution & Logistics Management, 25.19, 26-44. Senge, P. (1990). The Fifth Discipline. New York, NY: Doubleday Shim, J. K. & Siegel, J. G. (1999). Operations Management. New York, NY: Barron’s Educational Series. Simchi-Levi, D., Kaminsky, P., Simchi-Levi, E. (2002). Designing and managing the supply chain. 2nd ed. Boston, MS.: McGraw-Hill. Simon, H. A. (1947). Administrative Behavior. New York, NY: Macmillan Co. Slater, A. (1976). The significance of industrial logistics. International Journal of Physical Distribution, 7, 70-112. Smith, T. H. C., Srinivasan, V., & Thompson, G. L. (1977). Computational performance of three subtour elimination algorithms for solving asymmetric traveling salesman problems. Annals of Discrete Mathematics, 1, 495-506. Sölter, A. (1966). Die Kooperation dynamisiert den Wettbewerb. Industriekurier, 90, 6. Sölter, A. (1966a). Grundzüge industrieller Kooperationspolitik. Wirtschafts und Wettbewerb, 16, 223-262. Spence, M. & Zeckhauser, R. (1971). Insurance, information, and individual action. American Economic Review, 61.2, 380-387. SPIEGEL (2003, November 28). Mercedes-Chef Hubbert verteidigt deutsche Autos. Spiegel Online.

Retrieved

February

auto/aktuell/0,1518,275969,00.html

15,

2005,

from

http://www.spiegel.de/

290

References

Spreng, R.A., Mackenzie, S.B., & Olshavsky, R.W. (1996). A reexamination of the determinants of consumer satisfaction. Journal of Marketing, 60.3, 15-32. Stadtler, H.. (2000): Supply Chain Management – An Overview. In H. Stadtler. & C. Kilger (eds.), Supply Chain Management and Advanced Planning: Concepts, Models, Software and Case Studies (pp. 7-28). Berlin et al.: Springer. Stalk, G., Evans, P., & Shulman, L.E. (1992). Competing on capabilities: The new rules of corporate strategy. Harvard Business Review, 70.2, 57-69. Stank, T., Crum, M., & Arango, M. (1999). Benefits of interfirm coordination in food industry supply chains. Journal of Business Logistics, 20.2, 21-41. Stern, L. & El-Ansary, A. (1992). Marketing channels. 4th ed. Englewood Cliffs, NJ: Prentice Hall. Stevens, G. C. (1989). Integration of the supply chain. International Journal of Physical Distribution and Logistics Management, 19.8, 3-8. Stevenson, W. J. (2002). Operations Management. 7th ed. New York, NY: McGraw-Hill. Stölzle, W. & Otto, A. (2003). Supply Chain Controlling in Theorie und Praxis. Wiesbaden: Gabler. Subramani, M. (2004). How Do Suppliers Benefit from IT Use in Supply Chain Relationships. MIS Quarterly, 28.1, 45-74. Sugden, R. (1991). Rational choice: A survey of contributions from economics and philosophy. The Economic Journal, 101.407, 751-758. Sylver, E. A., Pyke, D. F., & Peterson, R. (1998). Inventory Management and Production Planning and Scheduling. 3rd ed. New York, NY, et al.: John Wiley & Sons. Tan, K. C. (2002). Supply chain management: Practices, concerns, and performance issues. Journal of Supply Chain Management, 38.1, 42-53. Tan, K. C., Kannan, V. R., & Handfield, R. B. (1998). Supply chain management: Supplier performance and firm performance. International Journal of Purchasing and Material Management, 34.3, 2-9.

References

291

Taylor, D. H. (1999). Measurement and analysis of demand amplification across the supply chain. The International Journal of Logistics Management, 10.2., 55-70. TCG

(2001).

Annual

Review

2001.

Retrieved

February

15,

2005,

from

http://www.tridencomponentsgroup.com/reports/2001%20Report.pdf TCG (n.d. a). TCG Unitech Systemtechnik – Austria. Retrieved February 15, 2005, from http://www.tridentcomponentsgroup.com/index.asp?page=site&site_id=17 TCG (n.d. b). TCG Unitech Systemtechnik – Austria. Retrieved February 15, 2005, from http://www.tridentcomponentsgroup.com/index.asp?page=site&site_id=17&display_order= 2 TCG (n.d. c). TCG Unitech Systemtechnik – Austria. Retrieved February 15, 2005, from http://www.tridentcomponentsgroup.com/index.asp?page=site&site_id=17&display_order= 3 TechTarget

(2003).

EDI.

Retrieved

February

15,

2005,

from

http://searchcio.techtarget.com/sDefinition/0,,sid19_gci213925,00.html Teigen, R. (1997). Information Flow in a Supply Chain Management System (Dissertation) .University of Toronto. Retrieved February 15, 2005, from http://www.eil.utoronto.ca /profiles/rune/dip-thesis.html Terreri, A. (2001). Profiting from cross docking. Warehousing Management, 8.8, 29-34. Thonemann, U. W. (2002). Improving supply-chain performance by sharing advance demand information. European Journal of Operational Research, 142.1, 81-107. Thorelli, H. B. (1986). Networks: Between markets and hierarchies. Strategic Management Journal, 7.1, 37-51. Timmermann, A. (1985). Strategisches Denken: Lebenslanges Lernen auch für Unternehmen. In H. Rafée & K. P. Wiedmann (eds.), Strategisches Marketing (pp. 197-227). Stuttgart: Poeschel. Tong, A. (2003). The rise--and fall?--of the keiretsu. ACN: Asian Chemical News, 9.385, 14. Toomey, J. W. (2000). Inventory Management: Principles, Concepts and Techniques. Norwell, MS: Kluwer Academic Publishers.

292

References

Towill, D. R. (1994). 1961 and all that: the influence of Jay Forrester and John Burbidge on the design of modern manufacturing systems. In Proceedings of the 1994 International Systems Dynamics Conference on Business Decision Making (pp. 105-115). Stirling, England. Towill, D. R. (1996). Industrial dynamics modeling of supply chains. International Journal of Physical Distribution and Logistics Management, 26.2, 23-42. Tröndle, D. (1987). Kooperationsmanagement: Steuerung interaktioneller Prozesse bei Unternehmenskooperationen. Bergisch-Gladbach et al.: Eul. Trunick, A. (1989). Outsourcing: A single source for many talents. Transportation and Distribution, 30.7, 20-23. VDA

(n.d.).

VDA

Abteilung

Logistik.

Retrieved

February

15,

2005,

from:

http://vda.de/de/vda/intern/organisation/abteilungen/logistic_02.html. Vitasek, K. (2002, October 02). Supply chain visions – logistics terms and glossary. Retrieved February 15, 2005, from http://www.clm1.org/resource/downloads/glossary100902. PDF. Volkswagen AG (n.d.). Geschäftsbericht 2002. Retrieved February 15, 2005, from http://volkswagen-ir.irkmm.de/deutsch/index.ie.html. Volkswagen Transport (n.d.). Volkswagen Transport. Retrieved February 15, 2005, from http://www.volkswagen-transport.de/englisch/index.html. von Neumann, J. (1951). Various techniques used in connection with random digits. In A. S. Householder, (ed.) The Monte Carlo Method (pp. 36-38). National Bureau of Standards Applied Mathematics Series, 12. Voronoi, M. G. (1908). Nouvelles applications des parametres continus à la theorie des formes quadratiques. Journal Reine Angewandte Mathematik, 134, 198-287. VW Group Supply.com (2003a). Prozessunterstützung durch EDI. Retrieved November 15, 2003,

from

http://www.vwgroupsupply.com/VWPortal/Navigation?ItemName=vw.

Zusammenarbeit.LogHilfe. VW Group Supply.com (2003b). Volume advantages. Retrieved November 15, 2003, from http://www.vwgroupsupply.com/ePortal/servlet/MenuActionServlet?MenuAction=Select&I temId=1861494833&MenuId=-642615006&State=vw.Home.HomeMenu;0;-

References

293

;ActiveRole;de,AT,-;-;-2058462299,-642615006&RS=vw.HomeHilfe.HomeHilfeMenu;0;;ActiveRole;de,AT,-;-;-1044216667,-1859845630&HS=vw.Home.HomeMenu;0;;ActiveRole;de,AT,-;-;-2058462299,-642615006&target=MultiPurposeFrame&CL=de,AT,&LS=vw.vwMenu;12;-;ActiveRole;de,AT,-;-1622935660;755338517. Wagner, M. (2001, February 28). BESI2/BESI3 – VVG – LAFES-Feinabruf. Retrieved November 15, 2003, from Audi Intranet. Wahler (n.d. a). Products and development solutions. Retrieved November 15, 2003, from: http://www.wahler.de/english/e_produkts.html. Wahler (n.d. b). Successful cooperation with the international car industry. Retrieved November 15, 2003, from: http://www.wahler.de/english/e_oe_partner.html. Wahler

(n.d.

c).

Locations.

Retrieved

November

15,

2003,

from

http://www.wahler.de/english/e_plants.html. Waller, D. L. (1999). Operations Management: A Supply Chain Approach. London et al.: International Thomson Publishing. Waller, M., Johnson, M. E., & Davis, T. (2002, April 30). Vendor management inventory in the retail

supply

chain.

Retrieved

February

15,

2005,

from

http://www.logisticcases.com/dc_cases/vmi_retail_sc.pdf Walter, F. (1975). Physical distribution: The concept of shared services (thinking ahead). Harvard Business Review, 53.2, 24-31. Walton, L. W. (1996). Partnership satisfaction: Using the underlying dimensions of supply chain partnership to measure current and expected levels of satisfaction. Journal of Business Logistics, 17.2, 57-75. Wassink, M. (2004, March 6): Deutsche Autos - Der Ruf leidet. Hamburger Abendblatt. Retrieved

February

15,

2005,

from

http://www.abendblatt.de/daten/2004/03

/06/270511.html Watters, J. K. & Biernacki, P. (1989). Targeted sampling: Options for the study of hidden populations. Social Problems, 36.4, 416-430.

294

References

Weber, M. (1922). Wirtschaft und Gesellschaft: Grundriß der verstehenden Soziologie. Tübingen: Mohr. Weitzel, T., Gellings, C., Beimborn, D., & König, W. (2003). IS Valuation Methods - Insights from Capital Markets Theory and Practice. In Proceedings of the 7th Pacific Asia Conference on Information Systems (PACIS 2003). July 2003, Adelaide, South Australia. Weitzel, T., Wendt, O., & von Westarp, F. (2000). Reconsidering Network Effect Theory. In Proceedings of the 8th European Conference on Information Systems (pp. 484-491). July 2000, Vienna, Austria. Whipple, J. S. & Gentry, J. J. (2000). A network comparison of alliance motives and achievements. Journal of Business & Industrial Marketing, 15.5, 301-322. Wholey, D. R., Bryce, C. L., & Sleeper, S. (2000). Organizations and efficiency. Retrieved February

15,

2005,

from

http://www.hsr.umn.edu/people/regular/wholey/

organizations_and_efficiency.pdf. Williamson, O. E. (1979). Transaction cost economics: The governance of contractual relations. Journal of Law and Economics, 22.2, 233-261. Williamson, O. E. (1981). The economics of organization: The transaction cost approach. American Journal of Sociology, 87.3, 548-577. Womack, J. P., Jones, D. T., & Roos, D. (1990). The machine that changed the world. New York, NY: Macmillan Publishing. Wüstner, E. (2005). Standardisierung und Konvertierung: Ökonomische Bewertung und Anwendung am Beispiel von XML/EDI. To appear in Aachen: Shaker. Wyner, G. M. & Malone, T. W. (1996). Cowboys or commanders: Does information technology lead to decentralization?. In J. I. DeGross, S. Jarvenpaa, & A. Srinivasan (eds.), Proceedings of the 17th International Conference on Information Systems (pp. 15-18). December 1996, Columbus, OH. Yellurkar, D. (2002). Successfully Managing Distributed Commerce. Montgomerie Research Inc. Retrieved February 15, 2005, from http://www.ascet.com/documents.asp? d_ID=1121. Visited on 08/22/2002.

References

295

Yin, R.K. (2002). Case Study Research-Design and Methods. 3rd ed. Newbury Park, CA: Sage. Yu, Z., Yan, H., & Cheng, T. C. E. (2001). Benefits of information sharing with supply chain partnerships. Industrial Management + Data Systems, 101.3/4, 114-119. Zäpfel, G. & Piekarz, B. (1996). Supply Chain Controlling. Vienna: Ueberreuter. Zhao, X., Xie, J., & Leung, J. (2002). The impact of forecasting model selection on the value of information sharing in a supply chain. European Journal of Operational Research, 142.2, 321344. Zeier, A. (2002). Ausdifferenzierung von Supply-Chain-Management-Standardsoftware in Richtung auf Betriebstypen und Branchen unter besonderer Berücksichtigung des SAP APO. Dettelbach: Röll. Zimmer, K. (2001). Koordination im Supply Chain Management. Wiesbaden: Gabler. Zinn, W. & Bowersox, D.J. (1988). Planning physical distribution with the principle of postponement. Journal of Business Logistics, 9.2, 117-136. Zinn, W. & Parasuraman, A. (1997). Scope and intensity of logistics-based strategic alliances. Industrial Markteting Mangement, 26.2, 137-147. 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. Zörgiebel, W. (1983). Technologie in der Wettbewerbsstrategie. Berlin: E. Schmidt.