Content uploaded by Francisco J Navarro-Meneses
Author content
All content in this area was uploaded by Francisco J Navarro-Meneses on Jun 06, 2016
Content may be subject to copyright.
FACULTAD DE CIENCIAS ECONÓMICAS, EMPRESARIALES Y TURISMO
DOCTORADO EN ECONOMÍA APLICADA
DEPARTAMENTO DE ECONOMÍA Y DIRECCIÓN DE EMPRESAS
COMPLEXITY-BASED VIEW OF THE FIRM:
THEORETICAL PERSPECTIVES,
METHODOLOGICAL FRAMEWORK AND
EMPIRICAL EVIDENCE
A Dissertation Presented
by
FRANCISCO JAVIER NAVARRO MENESES
Directed by Professors:
FEDERICO PABLO MARTÍ
JAVIER CARRILLO HERMOSILLA
Alcalá de Henares
March 2016
© Copyright by Francisco Javier Navarro Meneses, 2016
All Rights Reserved
LETTER OF ACCEPTANCE FROM THE THESIS DIRECTORS
LETTER OF ACCEPTANCE FROM THE DEPARTMENT DIRECTOR
This thesis is dedicated to my wife, María José, without whose unconditional
encouragement, enormous patience, hard work and dedication to our family, this thesis
would never have been written. Thank you.
***
To my beloved daughters Sofía and Alicia. Let this thesis be an inspiration for you
to believe in your dreams and cultivate a passion for learning.
***
To my parents, thank you for planting the seed of effort and abiding curiosity.
THIS PAGE INTENTIONALLY LEFT BLANK.
CONTENTS
ACKNOWLEDGEMENTS ...............................................................................................................xi
ABSTRACT ................................................................................................................................. xiii
LIST OF TABLES .......................................................................................................................... xv
LIST OF FIGURES....................................................................................................................... xvii
CHAPTER 1. INTRODUCTION ....................................................................................................... 1
CHAPTER 2. COMPLEXITY AND THE FIRM .................................................................................... 7
2.1 Introduction ....................................................................................................... 7
2.2 The Firm as a Complex System ......................................................................... 11
2.3 Complexity and Theory of the Firm .................................................................. 13
2.4 Literature on Complexity and the Firm ............................................................. 16
CHAPTER 3. BASIS FOR A NEW APPROACH ................................................................................ 21
3.1 Introduction ..................................................................................................... 21
3.2 Why a New Approach is Necessary? ................................................................. 22
3.3 Evidence for a Complexity-based Approach...................................................... 24
3.4 Conceptualization of the CBVF Approach ......................................................... 27
3.5 Refocus of the Theorizing Process .................................................................... 36
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM ................................................................ 45
4.1 Introduction ..................................................................................................... 45
4.2 Theoretical Constituents of the CBVF ............................................................... 48
4.3 Methodological Framework ............................................................................. 66
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH ......................... 157
5.1 Introduction ................................................................................................... 157
5.2 Field Research Design .................................................................................... 158
5.3 Results Obtained ............................................................................................ 170
5.4 Discussion and implications ........................................................................... 217
CONCLUSIONS ........................................................................................................................ 223
FURTHER RESEARCH ............................................................................................................... 231
APPENDIX A. REVIEW OF THEORIES OF THE FIRM .................................................................... 237
A.1 Introduction ................................................................................................... 237
A.2 Neoclassical Theory of the Firm ..................................................................... 238
CONTENTS
A.3 Transaction Costs Theory ............................................................................... 241
A.4 Agency Theory ............................................................................................... 243
A.5 Resource-based View of the Firm ................................................................... 246
A.6 Knowledge-based Theory of the Firm ............................................................. 250
A.7 Stakeholder Theory ........................................................................................ 252
A.8 Organizational Theories of the Firm ............................................................... 254
A.9 Developments in Strategy .............................................................................. 258
APPENDIX B. DELPHI’S EXPERT BENCHMARKING OF RESPONSES ............................................. 261
B.1 Interconnectedness between Constraints and Value Repositories ........................ 262
B.2 Interconnectedness among Value Repositories ..................................................... 263
B.3 Interconnectedness between Value Repositories and Operating Margin............... 265
APPENDIX C. PANEL MEMBERS ............................................................................................... 267
APPENDIX D. DELPHI QUESTIONNAIRES .................................................................................. 271
D.1 Delphi Round 1 Questionnaire ....................................................................... 271
D.2 Delphi Round 2 Questionnaire ....................................................................... 273
D.3 Delphi Round 3 Questionnaire ....................................................................... 275
D.4 Delphi Round 4 Questionnaire ....................................................................... 277
APPENDIX E. FIELD RESEARCH DATA........................................................................................ 279
E.1 Participants Profile (participants.xlsx) ............................................................ 279
E.2 Participation Data (participants.xlsx) .............................................................. 280
E.3 Delphi Round 1 Constraints (round_1_R.xlsx) ................................................. 284
E.4 Delphi Round 1 Value Repositories (round_1_R.xlsx) ..................................... 285
E.5 Delphi Round 2 Constraints (round_2_R.xlsx) ................................................. 287
E.6 Delphi Round 2 Value Repositories (round_2_R.xlsx) ..................................... 288
E.7 Delphi Rounds 3 and 4 Data ........................................................................... 289
E.8 Edges Values and Attributes (edgespanel.csv) ................................................ 290
E.9 Vertices Attributes (vertices.csv) .................................................................... 305
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH .............................................................. 307
F.1 Code for Delphi Method Analysis ................................................................... 307
F.2 Code for Network Analysis ............................................................................. 315
F.3 Code for Fuzzy Cognitive Map ........................................................................ 328
F.4 Code for Panel Members Benchmarking Dotplots .......................................... 331
CONTENTS
APPENDIX G. VALUEINAIRLINES.COM ...................................................................................... 333
APPENDIX H. SOFTWARE TOOLS .............................................................................................. 341
H.1 R Programming Software ............................................................................... 341
H.2 Network Workbench ...................................................................................... 342
H.3 Qualtrics Survey System ................................................................................. 343
BIBLIOGRAPHY ........................................................................................................................ 345
VITA ................................................................................................................................... 367
THIS PAGE INTENTIONALLY LEFT BLANK.
ACKNOWLEDGEMENTS
xi
ACKNOWLEDGEMENTS
This thesis has been an extraordinary journey of learning and discovery for me, indeed
one of the most rewarding knowledge experiences in my life. Now the memory of the
excitement associated with constructing theory and finding empirical support for it clearly
outweighs all the trouble and times of self-doubt. Notwithstanding, this my particular grain of
sand to social research would have not been possible without the help and support of many
people.
First and foremost, the author desires to express his grateful appreciation to Professors
Dr. Federico Pablo Martí and Dr. Javier Carrillo Hermosilla, under whose direction this thesis has
been brought to completion. They smoothed the path of my research by devoting much of their
valuable time to me. They provided advice and guidance to overcome the difficulties, and
challenged me to prove what started off being mere intuitions and beliefs from my own
professional experience. Thank you for your encouragement and motivation to accomplish this
“complex” work.
Thanks are also due to Dr. Elpiniki Papageorgiou, Assistant Professor at the
Technological Educational Institute of Central Greece, who kindly accepted to review the data
from the field research and generously offered her advice on the construction of a model of the
firm using Fuzzy Cognitive Maps, which served for testing purposes in the thesis.
ACKNOWLEDGEMENTS
xii
This thesis has also involved many highly experienced executives from the airlines
industry around the world, who took part, to a greater or lesser extent, in the field research
Experts’ Panel. Despite their busy agendas and responsibilities, they painstakingly sought to
address key issues related to our complexity-based approach in airlines, providing the author
with key data to test basic assumptions on the thesis. Without their insightful contributions and
enthusiasm for sharing their knowledge, the field research would not have been possible. They
are simply too many to name here. Thank you all.
In addition, I would like to thank the complexity community at online research platform
ResearchGate.net, where many scholars from everywhere provided me with valuable insights on
various intricate questions that I circulated. They not only helped me with recommendations on
how to better focus on difficult questions, but inspired me with their comments at my
beginnings and progressions in the fascinating world of complexity.
Finally, my acknowledgement would not be complete without expressing my gratitude
to the University of Alcala library services, which made available a good part of the bibliographic
resources that I needed to write this thesis.
ABSTRACT
xiii
ABSTRACT
The progress made by the theory of the firm has been outstanding in the last 80 years,
its central concepts having become foundational for any theoretical and practical work focused
on understanding the behavior of the firm. However, in spite of this remarkable achievement,
most of the firm’s real problems remain intractable. Furthermore, an increasing new class of
issues emerge which seemingly fall out of managers’ control, relentless challenging our current
theories and methods. The above are traits of what investigators call complexity, and a new
paradigm centered on tackling and managing it is being formed in the social sciences. To explore
the applicability, opportunities and consequences that this paradigm may have for increasing
our understanding of the firm, this thesis examines two distinct lines of inquiry: theoretical and
methodological. Building upon a novel theoretical characterization of the firm as a complex
system, whose fundamental units continuously create and exchange value, and a
comprehensive 4-stage methodological approach, both lines converge to form a robust
framework to practically grasp, envision and influence the behavior of the firm. Furthermore,
empirical evidence from a field research carried out at an international scale by the author on
the airline industry is provided that supports the applicability of our proposed complexity-based
view in practical areas such as scenario testing, planning and decision-making, using knowledge
from experts and Fuzzy Cognitive Mapping. Finally, given the extensive theoretical and practical
background covered by our approach and the limitations exposed, guidelines for further
ABSTRACT
xiv
research are provided in the thesis that should encourage the next rounds of investigators to dig
deeper and seize the yet unrealized opportunities offered by a complexity-based view of the
firm.
KEYWORDS: complexity, theory of the firm, field research, methodology, modeling.
LIST OF TABLES
xv
LIST OF TABLES
Table 1. Value Classification Technique ..................................................................................... 94
Table 2. Summary of the CBVF Methodological Framework .................................................... 155
Table 3. Delphi Survey, Distribution of Panel Members by Functional Area ............................. 171
Table 4. Distribution of Panel Members by Organic Position ................................................... 173
Table 5. Delphi Round 1: Summary of Top Cited Constraints ................................................... 176
Table 6. Delphi Round 1: Summary of Top Cited Value Repositories ........................................ 177
Table 7. Comparison of Top 10 Constraints, Round 1 vs Round 2 ............................................. 178
Table 8. Comparison of Top 15 Value Repositories, Round 1 vs Round 2 ................................. 180
Table 9. Most Important/Influential Value Repositories in the AVCN ....................................... 199
Table 10. AVCN-FCM Adjacency Matrix ................................................................................... 205
Table 11. AVCN-FCM Simulation Scenarios .............................................................................. 208
Table 12. Base Scenario Outcome State Vector ....................................................................... 209
Table 13. Scenario #1 Input Vector .......................................................................................... 210
Table 14. Scenario #1 Outcome State Vector ........................................................................... 211
Table 15. Scenario #2 Input Vector .......................................................................................... 213
LIST OF TABLES
xvi
Table 16. Scenario #2 Outcome State Vector ........................................................................... 214
Table 17. Scenario #3 Input Vector .......................................................................................... 215
Table 18. Scenario #3 Outcome State Vector ........................................................................... 216
Table 19. List of Panel members .............................................................................................. 269
Table 20. List of Non-disclosed Panel members ....................................................................... 269
LIST OF FIGURES
xvii
LIST OF FIGURES
Figure 1. Dimensions of Complexity of the Firm ........................................................................ 10
Figure 2. Illustration of the Multi-level Web of VRs ................................................................... 59
Figure 3. Stages of the CBVF Methodology ................................................................................ 73
Figure 4. Phased Approach to Resolve the Problem With Data .................................................. 77
Figure 5. The CBVF Methodology: Stage 1 Activities .................................................................. 82
Figure 6. The Multidimensional Value System Dynamics ........................................................... 84
Figure 7. The CBVF Method: Stage 2 Activities......................................................................... 109
Figure 8. Neuron with R Inputs ................................................................................................ 121
Figure 9. Example of How a SOM works .................................................................................. 128
Figure 10. Fundamental Connections in Bayesian Networks .................................................... 130
Figure 11. Causal Diagram of Management of a Forest ............................................................ 142
Figure 12. The Delphi Process ................................................................................................. 164
Figure 13. Distribution of Participants by Geography ............................................................... 172
Figure 14. Type of Responses Provided by Panel Members ..................................................... 174
Figure 15. Strength of Interconnectedness: Constraints-Value Repositories (Round 3) ............ 182
LIST OF FIGURES
xviii
Figure 16. Strength of Interconnectedness: Value Repositories (Round 3) ............................... 183
Figure 17. Strength of Interconnectedness: Value Repositories-Op. Margin (Round 3) ............ 184
Figure 18. Strength of Interconnectedness: Constraints-Value Repositories (Round 4) ............ 187
Figure 19. Strength of Interconnectedness: Value Repositories (Round 4) ............................... 188
Figure 20. Strength of Interconnectedness: Value Repositories-Op. Margin (Round 4) ............ 189
Figure 21. Airlines’ Value Creation Network (AVCN) ................................................................ 192
Figure 22. AVCN Pathfinder Network Scaling ........................................................................... 194
Figure 23. Clusters in the AVCN ............................................................................................... 196
Figure 24. AVCN Tree-like Dendrogram ................................................................................... 197
Figure 25. Interconnectedness of Value Repositories and Operating Margin ........................... 200
Figure 26. Base Scenario Inference Outcome .......................................................................... 209
Figure 27. Scenario #1 Inference Outcome .............................................................................. 211
Figure 28. Scenario #2 Inference Outcome .............................................................................. 213
Figure 29. Scenario #3 Inference Outcome .............................................................................. 216
Figure 30. Benchmarking of Responses: Constraints vs Value Repositories .............................. 263
Figure 31. Benchmarking of Responses: Value Repositories..................................................... 264
Figure 32. Benchmarking of Responses: Value Repositories vs Operating Margin .................... 265
CHAPTER 1. INTRODUCTION
1
CHAPTER 1.
INTRODUCTION
Much research endeavor has been devoted in the last 80 years to theorizing on the
behavior of the firm. Such has been the interest of scholars in this field that the number of
papers, articles, books and thesis in the subject is one of the largest among all the branches of
economics. As a matter of fact, the study of the firm as an intricate behavioral entity has
exponentially grown in last decades to become a fully consolidated stream of research, as
evidenced by the large community of dedicated investigators around the world and the various
theoretical developments and schools emerged.
In spite of this prolific output, this author submits that there is no certainty that we are
today more capable to explain the behavior of the firm than we were 80 years ago; much less to
envision or predict as to how a particular firm would behave under real-world circumstances.
Notwithstanding the foregoing, few theorists of the firm would usually react to the above
critique, partly because taking these critiques would seriously mean questioning fundamental
tenets of mainstream economics (Foss, Klein 2005).
The above consideration does not seek to waste or neglect the contributions made by
those involved in the advancement of the theory of the firm. Quite the opposite, this author
CHAPTER 1. INTRODUCTION
2
highlights as outstanding the achievements made by the theorists of the firm in helping
generations of scholars and practitioners better approach an understanding of the firm.
However, these same people increasingly encounter today a daunting class of problems
that current theories of the firm fail to explain, or for which its explanatory power remains
limited or subtle. By these problems we do not mean some extraordinary happenings in the life
of the firm, but events that are rather familiar to its everyday operations. We refer to problems
that generally involve a large number of interacting parts, transcend the conventional
boundaries of the firm, and impact its behavior in an unusual variety of ways. Moreover,
problems that are seemingly uncontrollable and unforeseeable, and often neglected because
they fall out of range for managers.
The above are signs of what researchers and practitioners call complexity, and this
author contends that most theories of the firm have remained somewhat oblivious of the need
to comprehend its significance and produce science-based insights into its sources and effects. It
is true that many theories of the firm have worked well at explaining disorganized complexity
―i.e. when the properties of the firm as a whole can be understood by using probability and
statistical methods (Weaver 1991)― but at the end, most of the firm’s real problems are left
intractable and unexplained.
Nevertheless, as our knowledge of reality grows and our analytical tools become more
powerful and sophisticated, many theorists of the firm have started to focus on developing new
conceptual and methodological frameworks to further explain firm’s organized complexity ―i.e.
the dynamic inter-relationships of the firm as a system, processes, and self-organization―, thus
shifting their attention towards new ideas, such as openness, emergent behavior, or dynamic
systems, and its collateral multivariate, non-linear, expert-based analytical tools.
CHAPTER 1. INTRODUCTION
3
Ultimately, these are traits suggesting that a new paradigm devoted to understanding,
predicting and influencing the behavior of complex systems is emerging out of our increased
awareness and capacity to tackle complexity (OECD 2008). Furthermore, this paradigm centered
on discovering and managing complexity has been born in the context of natural sciences ―i.e.
biology, physics, ecology, to name just a few― and has evolved over the past several decades to
spread into diverse fields in social sciences, where its particular potential for the study of the
firm remains mostly unrealized and full of possibilities.
This thesis is a research journey dedicated to dig deeper into the development of such
paradigm in the field of the theory of the firm. More specifically, this thesis explores i) the
theoretical implications derived from the application of the idea of complexity to the context of
the firm, ii) questions whether a complexity-based view of the firm can be brought to practice,
and iii) investigates the form in which this approach might be replicated. Furthermore, by
providing empirical evidence from real-life firms, the author tests in the thesis the feasibility of
a complexity-based view of the firm as a means to extend our understanding of the behavior of
the firm.
The underlying assumptions made in the thesis require that we address the problem of
complexity from a conceptual angle (Jaccard, Jacoby 2010), complementing old concepts with
novel ideas to address common difficult firm behavioral issues ―i.e. non-linearities, emergent
behavior, homeostasis, networks, or hierarchies. Therefore, by integrating old with new, our
complexity-based approach further enables researchers and practitioners to decompose the
firm complex structure and the relationships between its building blocks in a more realistic
fashion, and set them to interact with each other in a simulated environment. Hence the need
to put pieces together that past theories of the firm previously have been examining separately.
CHAPTER 1. INTRODUCTION
4
As we shall see, the above calls for the application of new qualitative research methods
(Coughlan, Coughlan 2002) that bridge the gap between theory, reality and practice, as to make
our complexity-based view of the firm usable in practical applications that give practitioners
some control of situations (Glaser, Strauss 2009). Eventually, our approach should bring new
opportunities for researchers to explore collective behavior and validate new theoretical
models, and for practitioners, to configure firm’s policy settings and try to anticipate
performance.
This thesis is organized into four main chapters, in addition to Introduction, Conclusions
and Further Research. Chapter 2 investigates the relations between complexity and the firm as a
complex system. Despite the difficulties to arrive at a universal definition of complexity, in this
chapter the author examines the different dimensions that complexity of the firm may have, and
concludes that the notion of “complexity” is an inherent feature of every firm operating in a
competitive environment and a key building block to take into consideration for further advance
of the theory of the firm.
Upon the theoretical and practical shortcomings identified by the author, Chapter 3 sets
the basis for a new approach and provides evidence as of why a reality bounded theory is
necessary. In particular, the author underlines the importance of a new conceptualization of the
firm, as well as the need to refocus the theorizing process around an idiosyncratic notion of
complexity of the firm.
Building on the above premises, Chapter 4 elaborates on the theoretical constituents of
our proposed complexity-based view of the firm, with an emphasis on the key features that
characterize the firm as an adaptive complex system. Moreover, as our complexity-based
approach is decidedly grounded in reality, its practical implementation requires a
CHAPTER 1. INTRODUCTION
5
methodological framework that guides researchers and practitioners on the practicalities of
addressing complexity. Therefore, Chapter 4 provides the basis for the design of such
methodology, specifically formulating a method comprised of four stages that covers the key
activities, tools and techniques that should serve to systematically untangle complexity of the
firm.
As no theory building process would be complete without empirically illustrating the
application of the theory, Chapter 5 presents the results of a field research carried out by the
author at an international scale in the airline industry. The results obtained give supporting
evidence of the applicability of the theory and method presented in this thesis, and of the
replicability of the key assumptions made to further advance the new paradigm of complexity in
the firm.
Finally, this thesis is complemented with eight appendices. Appendix A provides a
summary of the theoretical review carried out by the author on a selected number of seminal
theories of the firm, specifically focusing on the extent to which these theories address the
problem of complexity in the firm. Appendix B shows an example of an expert’s benchmarking of
responses, drafted and handed out to each and every member of the Experts’ Panel
participating in the thesis field research. Appendix C contains the list of members participating in
the Delphi Experts’ Panel. Appendix D provides a transcript of the questionnaires used across
the four stages of the Delphi process. Appendix E shows the data captured in the field research.
Appendix F contains the R programming code used in the field research. Together with the data
provided in Appendix E; this code should enable any interested investigator to reproduce the
outcomes obtained in the thesis. In Appendix G there are screenshots of the website
(http://www.valueinairlines.com) created and hosted by the author to serve as the main
CHAPTER 1. INTRODUCTION
6
communication tool with the members of the Experts’ Panel along the field research. Finally,
Appendix H includes a short description of the various software tools used in the thesis.
CHAPTER 2. COMPLEXITY AND THE FIRM
7
CHAPTER 2.
COMPLEXITY AND THE FIRM
2.1 Introduction
The Oxford English Dictionary defines “complexity” as something “consisting of parts or
elements not simply coordinated, but some of them involved in various degrees of subordination;
complicated, involved, intricate; not easily analysed or disentangled”. What this definition
invokes is that some of the terms used in the definition of “complexity” are subject to personal
consideration, or they are even context-sensitive. What is more, if we add to the above a review
of up to date literature on complexity, there seems to be no fixed or universal definition of
complexity. This suggests instead that the degree of intricacy, involvement, or complication of a
system depends entirely on the observer’s assessment and/or a context.
Furthermore, “complexity” is such a general concept that it means something different
to different people, even in the academic circles. As an example, the following list illustrates the
considerable disparity of what “complexity” entails to the complexity research community:
“Simply stated, complexity arises in situations where an increasing number of
independent variables begin interacting in interdependent and unpredictable ways”
(Sanders 2003).
CHAPTER 2. COMPLEXITY AND THE FIRM
8
“A great many quantities have been proposed as measures of something like
complexity. In fact, a variety of different measures would be required to capture all
our intuitive ideas about what is meant by complexity and by its opposite, simplicity”
(Gell-Mann 1995).
• “To us, complexity means that we have structure with variations” (Goldenfeld,
Kadanoff 1999).
• “In a general sense, the adjective “complex” describes a system or component that
by design or function or both is difficult to understand and verify. […] complexity is
determined by such factors as the number of components and the intricacy of the
interfaces between them the number and intricacy of conditional branches, the
degree of nesting, and the types of data structures” (Weng, Bhalla et al. 1999).
• “A complex system is literally one in which there are multiple interactions between
many different components” (Rind 1999).
• “Common to all studies of complexity are systems with multiple elements adapting
or re-acting to the pattern these elements create” (Arthur 1999).
• “Complexity starts when causality breaks down” (Editorial 2009).
Most of the foregoing statements characterize “complexity” as a structural feature of a
system, in such a way that complexity appears whenever a threshold limit is exceeded by the
quantity of components, the nature of the variations, or the number of interactions among
components. Thus, one may wonder where this threshold limit is set; and one possible answer
would be that this threshold plainly depends on the self-determined assumptions made by the
observer, or otherwise imposed by the context.
CHAPTER 2. COMPLEXITY AND THE FIRM
9
As pervasive as the idea of structural complexity may seem, it is not the only feature or
measure that we should consider when attempting to get a comprehensive idea of what
complexity is. We must consider other complementary measures, such as “environmental
complexity” and “process complexity”, the latter if we are specifically centered on dynamic
systems.
Environmental complexity takes into account the physical and metaphysical
environments where the firm operates. Conceptually, it addresses the rules and conditions from
within and outside the boundaries of the firm, which in turn determine the satisfaction of the
firm’s intentions or goals. To grasp a realistic idea of what environmental complexity is all about,
it is appropriate to measure various dimensions, including institutional complexity, geo-political
complexity, and competitive complexity.
Process complexity is a key measure of the difficulty of describing and executing a
process. Specifically, a “process” may be defined as an interactive algorithm used to execute
relational orders according to the rules and/or constraints set to satisfy a goal or intention
(Howard, Rolland et al. 2004). When a process is so-defined, it indirectly concerns to a system
that is led by an objective and organized by a relational order of its components, which is a
definition closely related to what the firm stands for.
Furthermore, process complexity addresses the execution of periodic and non-periodic
processes within the firm and in relation to its environment. An assessment of process
complexity would thus require measuring of relational complexity, action complexity and goal
complexity (Howard, Rolland et al. 2004). Note that process complexity is also a measure of the
evolution of the complexity of the firm, and the link between structural and environmental
complexity.
CHAPTER 2. COMPLEXITY AND THE FIRM
10
As described above, the notion of “complexity” is an inherent feature of every firm
operating in a competitive environment. As such, complexity may change in terms of magnitude,
i.e. some firms may present more complexity than others, but it will always be present in the
characterization of the firm as an open system. Fig.1 provides a graphical illustration of the
different dimensions of complexity described above.
Figure 1. Dimensions of Complexity of the Firm
Source: own elaboration
Complexity is, in short, not only the raison d'être of the complexity-based view of the
firm (CBVF) proposed in this thesis, but also the fundamental principle that this author suggests
should guide the search for new developments in the theory of the firm useful for explanatory
and predictive purposes, and for assisting practitioners in making decisions. For this purpose, as
CHAPTER 2. COMPLEXITY AND THE FIRM
11
we shall see next, the CBVF particularly calls for the integration of knowledge from other
scientific disciplines and the emergence of new idiosyncratic concepts and hybrid tools.
2.2 The Firm as a Complex System
A foundational consideration of our proposed CBVF, to bear in mind throughout this
thesis, is the characterization of the firm as an open system, embedded in a complex
environmental setting with which it continuously exchanges value.
From a practitioner perspective, this may be a rather intuitive idea, since the
interconnectedness between the firm and its outer environment appears both as commonplace
and as an apparent determinant of the behavior of the firm. Nevertheless, a closer look at some
of the most important theories of the firm (Appendix A) reveals a somewhat different state of
affairs, many of which aim at understanding the firm as the sum of the properties of its
individual components, without allowing much for the emergence of additional properties from
component interactions.
As of today, the reigning paradigm mostly consists of breaking up the firm into as many
different components as possible and studying the properties of the resulting parts —i.e.
transaction costs, property rights, agent relationships, stakeholders, knowledge. It is thought
that by doing so, we should then be able to understand the firm wholly, simply because the
parts constitute and determine the whole. This reductionist approach has seemingly provided a
powerful research methodology up until now, and become a guiding principle for acquiring
knowledge and a relatively consistent view of the firm’s life.
CHAPTER 2. COMPLEXITY AND THE FIRM
12
However, “dissecting” the firm to its slightest components and focusing on a single, or a
few, component/s, removes virtually all the properties associated with the real behavior of the
firm. The alternative offered by advocates of the behavioral theory of the firm to look at the
firm as a whole (Cyert, March 1963) is also limited. By focusing on the firm as a whole, we could
gather empirical data and explain the behavior of the firm in terms of phenomenological
models, the components of which would have no real significance.
We can agree that this reductionist approach has worked well for the last two or three
decades, illustrating situations in which its premises are valid and the descriptions provided are
useful (Kaneko 2006). But it is not yet clear that maintaining the same strategy in the twenty-
first century will suffice. Actually, there are many flaws in the theories of the firm that hamper
their explanatory power and, as a result, a new approach is all the more necessary if we are to
keep the theory of the firm alive and with practical relevance.
Yet, this is not a trivial task. To embed a higher level of complexity into the “core” of the
theory of the firm entails major consequences. The implications affect the way researchers build
theories and models of the firm and, even more, how they approach the theory of the firm
subject matter. Certainly, these consequences are deep and far reaching as set out below.
First, if we are to internalize complexity into the theory of the firm we should start by
changing our conceptualization of the firm. This fundamentally involves acknowledging that the
firm is an open complex system in continuous exchange of inputs and outputs with the
environment, and in continuous building up and breaking down of (its) components. Therefore,
a key assumption that every new approach to the firm needs to adopt is that the firm is not
going to behave as a closed system anymore and is never in equilibrium, but in a state of
continuous flow of change and exchange.
CHAPTER 2. COMPLEXITY AND THE FIRM
13
Second, to grasp (even a slight) understanding of the complex nature of the firm and of
the patterns of interaction between its components, the firm as a whole and its outer
environment, we not only need a new conceptualization of the firm, but to further develop a
more holistic complexity-based view of the firm. This new approach should be contingent on key
concepts, such as hierarchy, system boundaries, holism, network, synergism, etc. thoroughly
addressed in systems research and other non-economic complexity-centered disciplines (Kaneko
2006, Jørgensen 2006, Boogerd, Bruggeman et al. 2007, Alberghina, Westerhoff 2007).
Finally, a complexity-based view of the firm entails opening an entirely new set of
methodological perspectives, based on i) a more interdisciplinary philosophy that addresses the
behavior of the firm from simultaneously different knowledge dimensions and, ii) the use of
qualitative and quantitative hybrid tools specifically designed to tackle complexity —even if they
are to be borrowed from other scientific disciplines and were not original designed to solve
complex social or economic problems.
2.3 Complexity and Theory of the Firm
So far we have been using concepts such as “firm” and “theory of the firm”, which are
central for the development of our proposed complexity-based view of the firm. It is thus all the
more necessary that we stop for a moment and ask ourselves, What is that we call “firm”? What
do we exactly mean by “theory of the firm”? What scholars are actually doing when they
investigate on the theory of the firm? What is the scope and method(s) of such theory? Where
are the facts and evidence behind the theory?
CHAPTER 2. COMPLEXITY AND THE FIRM
14
Certainly, a number of questions arise when we attempt to recapitulate the key issues
under the consideration of the theory of the firm. In fact, the field of the theory of the firm is so
extensive that it is challenging to present the theory of the firm as a single, well-defined,
distinctive set of assumptions and methods, all of which neatly resemble a unique and round
theory of the firm. Quite the contrary, the theory of the firm is best examined as a broad body
of theoretical constructs ―i.e. organizational theory, strategy management, etc.―, each of
which aims at explaining the behavior of the firm from a different perspective and employs a
somewhat different methodology. Yet, all these different perspectives share the same subject of
research: the firm, and produce fragmented though complementary and interrelated knowledge
on the behavior of the firm.
At this point it is also useful to make a distinction between the economics of the firm
and the theory of the firm. The former concerns itself with issues related to the structure of the
firm, its organization and boundaries, whereas the latter usually refers to the analysis of the
behavior and strategies of the firm in certain market scenarios (Dietrich, Krafft 2012).
The author is well aware that the economic literature often characterizes the different
theories of the firm through their peculiar definitions of the “firm” and according to the
determination of its boundaries. These two issues have helped stress the diversity of theories of
the firm and highlighted the difficulties existing to place them under the same theoretical
umbrella. As for the purpose of this thesis, a firm may be a corporation, a group of corporations,
a part of a corporation, or it may be a partnership, or some combinations of all the above
(Papandreou 2000).
Given the large theoretical heterogeneity found, the author supports the idea to use a
broad conception of the theory of the firm in this thesis, mainly centered on the following
CHAPTER 2. COMPLEXITY AND THE FIRM
15
topics: i) the firm as a mechanism for the allocation of scarce resources, ii) the mechanisms for
decision-making within the firm, and iii) the organization of the firm (Mahoney 2005). More
particularly, the author will refer to the family of theories of the firm and its customary
methodologies, as “conventional” theories of the firm. This is meant to differentiate the
mainstream paradigms that hinge on premises and authors dating back to the mid-twentieth
century, from the multidimensional prone, non-linear, complexity-based view of the firm set out
later on this thesis.
For most of the so-called conventional theories of the firm (see Appendix A) the term
“complexity” appears more like the ability of the firm to interact with, or grasp resources from,
the outside world (market). Cyert et al. summarize this position clearly when they assert with
almost watchmaker's precision that “the information received from the market enables the firm
to apply its decision criterion, and the competitive system then proceeds to allocate resources
and produce output. The market information determines the behavior of the so- called firm”
(Cyert, Hedrick 1972). Consequently, in a greater or lesser degree within the set of conventional
theories of the firm, once the market conditions are described, the behavior of the firm is
“automatically” deduced from the assumptions, in a rather aprioristic and programmed-like
fashion.
From here on the author will use the term “complexity” in a broad sense to mean not
only the intricate number or type of relations held between the firm and the actors in the
environment —or the market where the exchange of resources takes place—, but also to refer
the emergence of the characteristics of an adaptive open complex system in the firm, where
components continuously interact with one another creating and exchanging value. Eventually,
CHAPTER 2. COMPLEXITY AND THE FIRM
16
“complexity” is used in this thesis to denote how realistically the postulates of the theory of the
firm approach the behavior of the firm.
2.4 Literature on Complexity and the Firm
A growing body of literature on complexity of the firm has developed in last years,
mainly linked to the fields of management and organizational studies and the attempts made to
approach organizations from a complexity perspective. The advancements achieved so far
reflect not only the scholars’ increasing interest on this topic, but also shows that a new theory
of the firm based on complexity might be at sight.
Complexity in management have got inspiration from concepts originating in disciplines
like chemistry, physics, biology, mathematics and computing. However, as many of the
complexity-related concepts are difficult to “translate” to the field of management, metaphors
are most often suggested. The downside of this approach is that it does not help much to
remove ambiguity, or make new conceptualizations based on complexity useful.
What follows are some of the key contributions that are shaping today’s research on
complexity in firm-related disciplines and which further support the case for our CBVF.
Allen et al., for example, have dedicated time to reflect upon how complexity science
has influenced management and organization studies over the past two decades. For them,
complexity science challenges not only the foundations of our knowledge, but also the
economic, political and social institutions we build upon that knowledge. In fact, the very idea of
viewing natural and social systems as complex adaptive ones constitutes a major revolution in
thinking which, according to these authors, “will have impacts on society as great as those of the
CHAPTER 2. COMPLEXITY AND THE FIRM
17
Enlightenment”. Furthermore, Allen et al. submit that adopting a complexity perspective has
important ontological, epistemological and axiological implications with which management
researchers and practitioners alike must come to terms. In sum, complexity science provides
scholars with a firm and scientifically anchored foundation from which to explore and
understand human organizations (Allen, Maguire et al. 2011).
In their comparative analysis of theoretical approaches to managing complexity in
organizations, Bohórquez and Espinosa echo the central role of self-organization in the life of
organizations. Their work attempts to identify the differences and similarities in the way the
notion of self-organization is explained in different theories of complex systems used in
management, and they end up grouping them as complex systems theories, complex adaptive
systems (CAS) and organizational cybernetics (Bohórquez, Espinosa 2015). The different
approaches have theoretical and methodological differences.
Gharajedaghi suggests that as the organization as a whole is becoming more
interdependent, the parts increasingly display choice and behave independently. This dilemma
thus requires a dual shift of paradigm: “the first shift results in the ability to see the organization
as a multi-minded, sociocultural system, a voluntary association of purposeful members who
have come together to serve themselves by serving a need in the environment. The second shift
helps us see through chaos and complexity and learn how to deal with an interdependent set of
variables” (Gharajedaghi 2011).
For Gorzeń-Mitka and Okręglicka, the “complex” view of reality is important in
understanding the activities of an organization, actually it is a natural consequence. Complexity
and uncertainty of the environment are key determinants for the search of new management
methods that fit in with the reality. However, despite the importance of complexity for
CHAPTER 2. COMPLEXITY AND THE FIRM
18
management, these authors contend that most companies have not introduced or implemented
yet a complexity management system/approach, or if they use one, they do not know whether it
is efficient and adequate. A literature review conducted by these authors shows that existing
complexity management strategies can be organized according to different management
approaches. Therefore, for each area of complexity –avoidance, reduction, transfer and control–
several strategies or complexity management models exist (Gorzeń-Mitka, Okręglicka 2015).
McMillan’s perspective is on management of change in all its rich complexity. What
complexity involves for organizations, she argues, is “acting differently and introducing change
using complexity-based principle” (McMillan 2008). On implementing a complexity approach,
management has to think differently about organizations and management, focusing on process
and dynamics and acknowledging the uncertainty, unpredictability and the paradoxical nature of
life in today’s organizations. The implications, according to this author, are evident: “a manager
considering change from a complexity standpoint will realize that it is pointless to attempt to
control all the key variables in a given situation and will instead focus on what it is possible to
know and understand” (McMillan 2008). She goes as far as to propose twelve principles for
introducing a complexity-based change process in organizations.
Stacey et al. are critical of the different ways in which complexity thinking is being taken
up by organizations. They understand organizations as complex responsive processes of relating,
and draw on the complexity sciences as a source of analogies, interpreting them through a
relationship psychology (Stacey, Griffin 2008). The authors show that complexity thinking
focuses attention on the emergence of genuine novelty in everyday processes of communicative
action through “the essentially responsive and participative nature of human processes of
relating and the radical unpredictability of their evolution” (Stacey, Griffin et al. 2000). According
CHAPTER 2. COMPLEXITY AND THE FIRM
19
to these authors, the complexity sciences can be brought together with psychology and
sociology in many different ways to form a whole spectrum of theories of human organization.
For Smith and Mitleton-Kelly (2011), complexity invokes organizations evolving in a
process that is systemic, emergent and context dependent. According to these authors,
successful organizations are those that promote feedback, self-organization and create a
learning environment, while at the same time become more tolerant and comfortable with
emergence, unpredictability and uncertainty (Smith, Mitleton-Kelly 2011).
THIS PAGE INTENTIONALLY LEFT BLANK.
CHAPTER 3. BASIS FOR A NEW APPROACH
21
CHAPTER 3.
BASIS FOR A NEW APPROACH
3.1 Introduction
A review on some of the most prominent theories of the firm (Appendix A) reveals that
though they mostly provide the fundamental “building stones” for understanding the behavior
of the firm, there is still much room for enhancing their explanatory power in the form of more
realistic constructs.
Years of outstanding theoretical and methodological advancements have led us to what
we know today about the firm, and how we explain its behavior. In our continuous strive for
improvement, questions such as the increasing attention paid by scholars to the role of the
environment in the behavior of the firm are an example that illustrates how most recent
developments in the theory of the firm are starting to overcome some of its main limitations.
The field of theories of the firm have unquestionably made a quality leap in most recent
times, adding more “complexity” to their assumptions, which in turn points the way ahead for
making more practical progress. From this trend, the author suggests that a new theory of the
firm, centered on the complexity of the firm, might be at sight.
CHAPTER 3. BASIS FOR A NEW APPROACH
22
In this chapter the author provides further arguments according to which a new
approach to the behavior of the firm seems inevitable, thereby setting the basis for a new
conceptualization and theorizing process of the theory of the firm.
3.2 Why a New Approach is Necessary?
The question of why a new approach is necessary might be best answered with a
statement of principles: a new approach is necessary because conventional theories of the firm
—with their mostly linear, equilibrium-based, cause and effect thinking, and yet to improve
empirical evidence—, largely deviates from the real-business life, where firms operate and
develop.
It is precisely the author’s interest in the flaws evidenced by the conventional theories
of the firm to explain the real world and the particular chunks of reality that they ignore, which
point to the need for making new improvements in the theory of the firm. As a matter of fact,
there are many arguments at play that support this position.
We argue that there is a strong case for challenging the validity of some fundamental
assumptions made by the conventional theories of the firm, particularly those concerned with
propositions such as the firm as a profit-maximizing agent, how firms get started, the role of the
entrepreneur in a modern XXI-st century’s society (Papandreou 2000), or the way the firm
understands value creation. All of them need improvement, since they do not adequately reflect
the way actual firms operate.
Another much controversial issue arises from the study of the boundaries of the firm.
The discussion on the boundaries of the firm has been a hot topic in the field for years, and one
CHAPTER 3. BASIS FOR A NEW APPROACH
23
that has been useful for analyzing the various historical forms in which economic production and
distribution have been organized and coordinated. Over the past century, our notion of
boundaries has experienced significant modifications that reflect changes in the way firms have
been conceptualized (Martin 2012).
The two strains of work that have dominated the research on the boundaries of the
firm, transaction cost and property right economics, are arguably no longer valid to support the
significant organizational changes that are taking place (Holmström, Roberts 1998). For some
authors, the question of understanding where the boundaries of a firm lie is even a matter of
interpretation and “managers need to consider alternative interpretations which they and others
might make in any specific situation as this is a question subject to interpretation” (Blois 2006).
Whatever the case may be, the very concept of boundaries has become blurred as result
of the changes in the environment of business activities and the more open networked
formations that are taking place, both within and between firms (Nolan 2007, Cantwell 2013).
Reality is stubborn, and constantly remind us that firms are complex entities that deal with a
much richer variety of problems than conventional theories can withstand.
But there are still more arguments to consider. The conspicuous lack of a practical focus
evidenced by the most salient theories of the firm has led to serious utilization problems. By
this we do not mean that the theories of the firm must be devoted to the application of
scientific knowledge to solve specific managerial problems —which would be the domain of
practitioners—, but the development of theoretical knowledge that enables managers to solve a
class of managerial problems (van Aken 2004). Perhaps not intentionally, the gap created
between the theories of the firm and practical reality has greatly diminished their practical
relevance.
CHAPTER 3. BASIS FOR A NEW APPROACH
24
3.3 Evidence for a Complexity-based Approach
The approach to the firm as an open complex system is not something new. In fact, the
theorizing effort in the last decades on disciplines related to the theory of the firm ―i.e.
organizational theory, management studies, behavioral economics, social systems, or consumer
psychology― is a good example of the aspiration of the research community to explain the
behavior of the firm with more detail and precision, though from different theoretical
perspectives and with different purposes in mind.
As threshed as the field of the theory of the firm seems can be, why, then, do we
suggest a new “view” or approach to the firm? Why is it worth to dedicate more time and brains
to this seasoned topic of the theory of the firm, which has produced by the way, hundreds of
thoughtful papers and research literature? What else can a complexity-based view of the firm
add to our knowledge that we cannot obtain by mixing some of our currently available
conventional approaches? The facts are that much evidence suggests that the conventional
theories of the firm simply fail to tackle the key challenges posed by real business life of the
twenty-first century, and that a complexity-based view of the firm might be a plausible
alternative.
As Sanders (2003) submits: “the challenges we face today and those we’ll confront in the
future require new ways of thinking about and understanding the complex, interconnected and
rapidly changing world in which we live and work, and insights arising from the study of complex
systems are helping us expand our thinking in new directions”. Building upon this idea, we set
forth below what we consider are key evidences supporting the need for our CBVF.
CHAPTER 3. BASIS FOR A NEW APPROACH
25
First, it should be noted that significant progress has been made in complexity-related
studies within a broad range of non-economic scientific disciplines —e.g. theoretical physics,
biology, ecology, software computing. This has inevitably pushed forward their respective
frontiers of knowledge, building upon innovative and idiosyncratic methods aimed at tackling
complexity. No doubt such precedents are an excellent source of methods and tools —as well as
inspirational ideas— for the introduction of higher levels of complexity in the general economic
theory and, particularly, in the theory of the firm. Add to this picture the fact that never before
in the history of economics —and of social sciences— have researchers had the tools and
techniques to come face to face with complexity as they have today, hence the chance to seize
the opportunities provided by a new complexity-based view of the firm.
Second, a growing number of scholars in the field of general economic theory and in the
theory of the firm openly admit that economic life does not obey laws expressed in terms of
For years we have realized that the real world of the firm is not describable merely in
terms of prices, production and costs, and that the firm does not function in a mechanical way.
Instead, we have learned, in one way or another, that the real firm might function like a
mechanism, but more like a complex system than a precision machine. This analogy has been
noted by some authors, though not yet developed sufficiently to seize all its theoretical and
practical possibilities (Jaynes 1991).
What all the above fundamentally means is that our problem is not just one of
quantitative nature. Our theories are qualitatively wrong (Jaynes 1991) or incomplete, either
because we have chosen the wrong set of variables, or because our techniques are not
sufficiently sophisticated to grasp its key relationships and interactions. At this point, no
CHAPTER 3. BASIS FOR A NEW APPROACH
26
mathematical model or computing power can help us. It may be a matter that we have not yet
fully understood what complexity entails for the economic system and the firm.
Third, closely linked to the previous evidences is the acceptance of the premise that
every component of the firm is related to the whole system and vice versa, and therefore that
no meaning can be gathered out if it is excluded from the whole (Karsten 1990). In other words,
the firm can no longer be interpreted as a self-contained structure and examined with reference
solely to its constituent parts, but instead it forms a holistic construct. In this respect, we could
use a metaphor from modern physics to explain the whole/component relationship of our
proposed CBVF, saying that the universe is viewed as a dynamic part of an inseparable whole
and where the conventional concepts of space and time, and of cause and effect, have lost their
(conventional) meaning (Capra 1985).
Fourth, the firm research community seems today more prone than ever to changing its
mood towards complexity. This is reflected by how complexity is no longer considered the dead
end where most investigations come to a halt, or where theories of the firm lose their grip, but
the starting point for many innovative approaches. No doubt that the term complexity is not as
dreaded today as it was years ago, and that a growing number of scholars is now starting to
acknowledge complexity as the basis on which to build their research. This suggests that by
embracing a CBVF-like approach, new opportunities for the study of the firm should arise, which
might in turn greatly boost the theoretical and practical outcomes.
Last but not least, by embedding complexity into the “core” of the theory of the firm,
we are keeping a closer eye on reality and showing commitment to generate practical outcomes
that help practitioners and society in general —especially at a time when they increasingly
demand more and better practical results out of the researchers’ work. As argued above, the
CHAPTER 3. BASIS FOR A NEW APPROACH
27
CBVF approach is seriously determined to keep the theory of the firm bounded to reality, which
in turn involves producing practical theories and fulfil its obligations with society with regards to
general progress and wellbeing.
3.4 Conceptualization of the CBVF Approach
Future theoretical and methodological developments in the theory of the firm will
necessarily depend upon coming to grips with complexity and open adaptive systems. Such
approaches involve a distinct view of the firm that is complex both in its composition —many
components interacting with each other and with their environment on multiple levels— and in
the rich diversity of behavior of which they are capable. The following section provides a general
overview of the key conceptual elements necessary to articulate and realize future
developments in the theory of the firm when complexity is introduced.
3.4.1 Open complex system approach
The adoption of an open complex systems approach in the theory of the firm is nothing
new. Its philosophical roots can be traced back to the discussions between advocates of the
mechanistic and organismic models of the 19th and early 20th centuries (Kast, Rosenzweig
1972, Johnson, Kast et al. 1973), and can be epitomized when Scott asserts that “the only
meaningful way to study organization is to study it as a system (…) Modern organization theory
and general system theory are similar in that they look at organization as an integrated whole”
(Scott 1961). In fact, he was not in solitude, Cyert and March also note that organizations are
“complex systems” and they went as far as to develop a model comprised of a set of
CHAPTER 3. BASIS FOR A NEW APPROACH
28
interdependent decision rules, responding to both external feedback and to internal
reinforcement (Gavetti, Greve et al. 2012).
Not surprisingly, many firm’s practitioners in the last century have been using a systems
approach intuitively (and implicitly), without even knowing much about the underlying stream
of systems research. As a matter of fact, all that practitioners needed to do is develop an
intuitive sense of the situation of the firm, act as if they were flexible diagnosticians, and adjust
their actions and decisions accordingly (Kast, Rosenzweig 1972).
Today, the open systems approach is touted as a promising means to better understand
the complexity of “live” organizations, and as Simon observes “its popularity is more a response
to a pressing need for synthesizing and analyzing complexity than it is to any large development
of a body of knowledge and technique for dealing with complexity”(Simon 1962a).
The key idea behind the open system approach is that firms are themselves open
systems, maintaining in continuous exchange of inputs and outputs with the outer environment,
and in continuous building up and breaking down of their components (Von Bertalanffy 1950).
This hardly novel but powerful idea comes to break with the approaches used in conventional
theories of the firm, centered almost exclusively on closed or semi-closed systems and fictitious
equilibria. Instead, according to the open systems approach, real business life shows that the
firm as a whole does not behave as a closed system and is never in equilibrium, but instead in a
state of continuous flow of change.
The implications of the open systems approach for the theory of the firm are broad and
varied. One of the main concerns has to do with the way in which the firm responds to
environmentally generated inputs, and how it adapts internally to these environmental forces.
CHAPTER 3. BASIS FOR A NEW APPROACH
29
This fundamentally leads us to a new conceptual model, contingent on concepts such as holism,
synergism, hierarchy, system boundaries, dynamic equilibrium, multiple goal-seeking, to name
just a few, and the need to define more explicitly certain patterns of relationships and
interactions between the firm, its components, and key environmental variables.
As a result of the above considerations, it becomes pressing for the complexity-based
approaches to the theory of the firm to extend their assumptions far beyond the boundaries of
the firm, and transcend thus far isolated economic disciplines. This shall necessarily involve
establishing a broader formulation of the theory of the firm, which comprises states of non-
equilibrium as well as those of equilibrium.
In short, it would be highly desirable that the complexity-based approach attempt to
“refound” the fundamental underpinnings of the theory of the firm, and incorporate the key
principles governing open adaptive systems. Such a consideration of the firm as an open system
would most probably lead to relevant quantitative and qualitative new insights, what would
itself further increase our abilities to understand the real behavior of the firm.
The author is aware that although the open system approach has been broadly
developed in most of the so-called “hard sciences”, further analysis remains to be done on the
applicability of open systems to social sciences and, more specifically, to the behavior of the
firm. In pursuing this endeavor there will most likely appear difficulties that stem from the
novelty of the open system paradigm, and our inability to operationalize “all we think we know”
about open systems. It is thus a challenge for the research community to make the open system
approach more explicit and try to merge it into existing firm research trends.
CHAPTER 3. BASIS FOR A NEW APPROACH
30
Only the time and commitment of the firm research community in their struggle to find
new answers to complex behavioral questions will tell how successful the open systems
approach can be. After all, it should not be forgotten that one of the major contributions made
by the introduction of an open systems mindset is to prevent us from accepting as final a level of
theoretical analysis that is below the level of the empirical world we are investigating (Boulding
1956). The open systems approach might well spur that step forward in the scale of complexity
that the theory of the firm inexorably deserves.
3.4.2 Interdisciplinarity
There is growing awareness in the research and practitioners communities of the need
to accomplish the study of the firm leaving aside the (theoretical and methodological) shortcuts
that hamper our ability to understand the behavior of the firm from an interdisciplinary
perspective. This call for new means to better understand the behavior of the firm is also
reflected in the works of many theorists of the firm, who have long requested a more
interdisciplinary approach —see for example the papers published in the Journal of
Interdisciplinary Economics.
When pursuing an interdisciplinary approach, one of the most important tasks to
accomplish is transforming the theory of the firm into a truly hybrid discipline. This basically
involves acknowledging that the economic theory alone is not enough to explain firm’s behavior,
and that we need to gain input from diverse fields of knowledge —such as from sociology,
biology, physics, software computing, anthropology, psychology, etc.— to grasp a higher degree
of understanding.
CHAPTER 3. BASIS FOR A NEW APPROACH
31
A valuable starting framework of reference for the implementation of an
interdisciplinary approach within the theory of the firm is provided by the General Systems
Theory (GST). GST aims at providing scientists with an integrated frame of knowledge so that
they can understand and communicate with each other —i.e. GST would enable firm
researchers to find out new answers to key questions by interacting with scientists in related
areas of knowledge such as computing scientists, data scientists, sociologists, biologists, etc.
(Boulding 1956). GST also supports the idea of isomorphism, which should enable us to identify
and make use of common elements from within the scientific universe, as they are all basically
concerned with the same phenomena.
The space of opportunities for improvement of the theory of the firm may be
substantial, shall we apply a more interdisciplinary approach. In fact, prominent authors like
Boulding (1956) specifically cited management science as a field of opportunity to break with old
mechanistic methods and foster more powerful and fruitful approaches.
Nevertheless, the advocates of GST are not the only ones pressing for the
interdisciplinary approach. Other prominent authors such as Edith Penrose, have also claimed
that under complexity and diversity a firm could be approached with many different types of
analysis —i.e. sociological, organizational, engineering, economic— as well as from whatever
standpoint that seems appropriate to the firm’s concrete problem at hand (Penrose 1959).
Simon (1982) even submits that “there seems to be no escape from psychology”, and he
goes as far as saying that as organizational economics and strategic management deal with
uncertainty, “they will have to understand how humans in fact behave in the face of uncertainty,
and by what limits of information and computability humans are bound”. Furthermore, in
CHAPTER 3. BASIS FOR A NEW APPROACH
32
asserting that organizational economics and strategic management are like chess, inevitably
culture-bounded and history-bounded, Simon stretches the interdisciplinary focus.
An example of how the interdisciplinary approach may become a key concept for the
development of complexity-based theories of the firm is illustrated by the growing importance
given to terms like “trust” when it comes to describe the behavior of the firm. As Arrow (1974)
denotes, trust is an “important lubricant of a social system” and, together with other values
such as loyalty and authenticity, portrays a positive externality that cannot be analyzed from a
strict economic perspective to become fully understandable, but only through an
interdisciplinary approach.
3.4.3 Empirical new modeling tools
A predominant number of investigations carried out under the conventional theories of
the firm have involved sample studies using secondary data, most often from public business
databases. Standard multiple regression models are the dominant statistical technique, together
with the use of correlations and analyses of variance (ANOVAs). In some cases, studies have
been performed using a cross-sectional design with static specifications of the relationships
under examination, though most studies do not include an effective set of control variables
(Hitt, Gimeno et al. 1998).
What the above means is that most conventional theories of the firm have a limited
perspective of the big picture. Researchers frequently open their minds to those inputs which
they can handle within their bag of tools and they often dismiss variables outside their interest
or competence as being irrelevant. As Kast et al. (1972) observe, “we are hampered because
CHAPTER 3. BASIS FOR A NEW APPROACH
33
each of the academic disciplines has taken a narrow “partial systems view” and find comfort in
the relative certainty which this creates”.
However, as research in the theory of the firm continues to develop —and other related
disciplines, such as organizations and strategic management— better research, refined theory,
and more powerful and sophisticated analytical methods have entered the arena. New and
better sources of data have also been developed, along with a better understanding of the field
investigated, all of which is resulting in more and better outcomes (Hitt, Gimeno et al. 1998).
But despite the progress made, the shift from the conventional conceptions of the firm
to one consisting of a complex open system entails an entirely new scientific style and use of a
full new set of advanced modeling tools. From an emphasis on deductive reasoning within a
tight system of axioms, the complexity-based theories of the firm demand that higher attention
is paid to a detailed empirical exploration of complex algorithms of thought (Mahoney 2005)
and multivariate phenomena, if the goal is to keep the theory relevant.
The classical world of magnitudes thus needs to give way to the analysis of interactions
between the different components and subsystems of the firm, and between these and
environmental interfaces. Interactions are now key, and researchers need to measure them
proficiently.
Along this process, many challenges will arise in complex firm modeling that threaten
our research integrity. One of the major challenges is that the need to deal with comprehensive
systems of relationships overruns our ability to fully understand and predict the firm
interactions (Kast, Rosenzweig 1972). Furthermore, most of the current methods and tools used
by social scientists may not be sophisticated enough to grasp the relationships among
CHAPTER 3. BASIS FOR A NEW APPROACH
34
components and subsystems, not to say to gather the informational inputs that are necessary to
make the systems approach really work. It is worth noting that only today we are beginning to
understand multivariable relationships, at the expense of pushing the limits of our capacities.
Consequently, if the theory of the firm is to advance and make contributions of practical
relevance, it must either adopt new empirically-oriented modeling tools from outside the
discipline, or develop its own.
In this regard, the forecast made by Hitt et al. (1998) on the future tools and methods
that strategic management research will use, seem highly applicable in the field of the theory of
the firm today. In particular, these authors forecast that future strategic management research
will use: (a) methods appropriate for longitudinal or panel samples, (b) explicitly dynamic
analytical methods, (c) methods appropriate for studying discrete strategic choices, behaviors,
or actions, (d) methods that acknowledge the interdependence of firms with other firms or
actors in their environment, (e) methods that explicitly account for the heterogeneity of firms,
(f) methods that uncover the causal structure among and the endogeneity of variables, and (g)
methods that account for the imperfect measurement of strategic constructs.
As we would expect, many of these methods and its related tools will have to be
borrowed from other scientific disciplines, such as computer science, systems biology,
neuroscience, eco-modeling, structural sociology, marketing, psychology, among others. Note
that although the methods described above emphasize quantitative methodologies, an
important challenge remains ahead for the complexity-based theories of the firm to integrate
qualitative and quantitative research and make use of non-conventional research methods.
CHAPTER 3. BASIS FOR A NEW APPROACH
35
3.4.4 Extensive management of information
When carrying out the adoption of the concepts above, information becomes a critical
asset. Notwithstanding the grounds for this view is nothing new. The difference today lies in that
firms are embedded in environments of varying complexity, thus information impacts the
behavior of the firm at a higher rate, and to a deeper extent. In fact, no firm can prosper today
unless it is able to articulate knowledge from within the organization and its environment, and
to permeate all levels of the organization and drive it to action.
Information is not only key for adjusting internal decision-making procedures that
account for environmental variations, but it is also a factor required to create strong
relationships and coherent organizational architectures. Consequently, the integration of
information management into the new theoretical and practical perspectives of the firm is a
necessary condition to improve the scope and depth of the new (complexity-based) theories of
the firm.
The next generation of theories of the firm will thus give priority to the processes by
which the firm gathers, operates and makes information actionable. This fundamentally
concerns the firm’s maturity cycle of information, from the point where the information is
captured, to the point where it is turned into usable knowledge and disseminated within and
outside the boundaries of the firm.
An example of the impact that the management of information has over the behavior of
the firm is the level of attention reached by the search and transmission of information,
currently considered a vital step in the decision-making process of every firm. It is hardly
CHAPTER 3. BASIS FOR A NEW APPROACH
36
surprising that the emphasis given to how the firm acquires and processes information is one
way in which “complexity” has been introduced into the theory of the firm.
3.5 Refocus of the Theorizing Process
Theorizing from a complexity standpoint entails the need to deconstruct the very own
hierarchical structure of the firm. To apprehend the concept of hierarchy, we resort to Simon’s
(1962) definition, which states that a system is composed of interrelated subsystems (or
components), each of which is, in turn, hierarchic in structure until it reaches some lowest level
of elementary subsystem. The idea of hierarchy greatly simplifies the description of a complex
system and, as in the case of the firm, facilitates the understanding of how it behaves.
It is worth noting that considerable discussion has taken place in the research
community as to what comes first, whether our vision of the world as a hierarchical system, or
our need to use the idea of hierarchy as the only way to get useful knowledge about a complex
system (Simon 1962). Of course this is not the place to go deeper into this controversy, although
the debate itself denotes the extent to which the term “hierarchy” is a key concept to consider
in the study of complexity of the firm.
The idea of the firm as a hierarchical system is thus central for the development of our
complexity-based view of the firm. Furthermore, we could hardly understand, describe, or
analyze the firm as a whole or its interrelated constituent parts, unless we are able to unravel
the hierarchy of the firm. Without such notion at hand, the complexity of the firm will plainly
exceed our information processing capacity and our own ability to generate any practical
CHAPTER 3. BASIS FOR A NEW APPROACH
37
understanding of the firm’s behavior. The latter resulting in either an incomplete, or a scarcely
useful theory of the firm.
3.5.1 Unraveling of the hierarchy of the firm
The first obstacle that emerges when we attempt to unravel the hierarchy of the firm is
to figure out what is the basic building block of the firm; in other words, what is the
fundamental piece or “atomic particle” to which the remaining components of the firm can be
reduced. For example, in astronomy the basic building blocks are the stars or planets; in biology,
the cell or proteins (Simon 1962a).
Within our approach of the firm as an open adaptive system, the fundamental building
block that we will consider, from which the firm’s components (and subsystems) become
differentiated and cohesive, is “value”. Value is the brick and mortar that gives consistency to
the firm’s hierarchical structure. Without creating and exchanging value, the firm’s components
would collapse inevitably, and the firm would cease to exist.
Notwithstanding the foregoing, the term “value” is a controversial concept that means
different things to different people. Moreover, most of the research literature on “value” comes
from neoclassical bounded economic theory, and there is little consistency in the approaches.
The term “value” not only invokes an economic game between the firm’s components,
and between these and external agents, but a perceived preference for a particular way of doing
things —i.e. the products and services that a component of the firm produces—, which in turn
facilitates (or blocks) their goals. Hence, we can recap as follows: the value created by the firm
entails an attitude toward, or an emotional bond with the firm (or its components), comprising
CHAPTER 3. BASIS FOR A NEW APPROACH
38
an interactive, dynamic, and contextual preference and experience (Brandenburger, Stuart
1996).
Since no universal meaning of “value” seems plausible, in the thesis we will use this
term not in a strict economic sense —i.e. the difference between the willingness-to-pay of a
buyer minus the opportunity cost of the supplier (Brandenburger, Stuart 1996)— but in a much
broader sense. More specifically, we will refer to the firm’s Multidimensional Value System
Dynamics (see Section 4.3.5, B.1), which comprises both the internal and external value creation
and exchange processes flowing to and from the components of the firm.
Upon embracing the idea of “value” as the basic building block of the firm-system, the
next step leads us to substantiate the firm’s fundamental components as “value repositories”.
Value repositories exhibit a duality between value creation and exchange or, in other words,
between the stock of distinctive value accumulated by the firm ―generally in the form of
Potential Use Value― and the dynamic realization of that value with a customer ―in the form of
Exchange Value (Bowman, Ambrosini 2000, Lepak, Smith et al. 2007).
Value repositories have an interconnected nature, thus forming a network that links
them together at different levels of hierarchy and strength. For example, as we shall see in
Chapter 5, the network of most important value repositories in an airline is made of 15 unique
and highly connected value repositories.
An important consideration to bear in mind is that value repositories are different from
most firms’ organizational units, as the value creation processes usually require inputs from
different sources, other than a particular well-delimited organizational unit. For example, the
“Process and cost optimization” value repository pinpointed in our airlines field research
CHAPTER 3. BASIS FOR A NEW APPROACH
39
(Chapter 5), does not match with any particular organizational unit in an real life airline. Instead,
this value repository features the distinctive value created and exchanged by airlines ―allegedly
involving several organizational units― aimed at simplifying the relationship with the customers
and gain internal efficiency.
To effectively unravel the hierarchy of the firm, we shall thus assess the firm’s value
system dynamics through the interactions taking place between value repositories. This involves
mapping the strength of the interactions, its temporality, and where and with whom the
strongest and weakest interactions happen.
In this process it is important to realize that only a small fraction of the interactions
between value repositories will have real influence on the behavior of the firm. At this point it is
useful to bring up Simon’s notion of “redundancy” (Simon 1962a), which states that the
underlying complexity of a system can be reduced to only a limited amount of subsystems and
interactions that are truly relevant and differentiated.
The idea of redundancy shall enable us to limit the number of components (value
repositories) in which we should focus to only a fraction, provided that we have accomplished
an iterative assessment process that discerns between strong and weak interactions (Simon
1976). Once we get to know the strongest fraction of all possible interactions, we will be closer
to achieve a better representation of the behavior of the firm —as the remaining interactions
are most likely weak and not so relevant for our analysis of complexity.
CHAPTER 3. BASIS FOR A NEW APPROACH
40
3.5.2 Decomposition of the interactions among firm’s components
Let us first explain what we mean by “interactions”, and how other scientific disciplines
use this term.
Usually when molecular biologists say that cells interact, what they are most probably
thinking is that some effect is communicated between cells through receptors, or the exchange
of certain molecules. Ecologists most likely would consider “interaction” as the relation of
individuals in an ecological system such as prey–predator relationship. Physicists and chemists
would consider interaction to be the influences communicated by forces –intermolecular,
electromagnetic, etc. (Kaneko 2006).
For the purpose of this thesis, we use the term “interaction” in a broad sense, meaning
the mutual effects resulting from exchanges among the firm’s value repositories. For example,
we regard as an interaction the exchange of value flowing from value repository A to value
repository B, or the redistribution of resources between two value repositories.
The interactions between the firm’s value repositories can be decomposed according to
the following set of variables:
Intensity. The intensity or strength of the interactions between value repositories
provide us an idea of the firm’s different levels of hierarchy. Some interactions may
be labeled as weak, whereas others can be strong. In order to identify the firm’s
levels of hierarchy, we must first examine with whom do the value repositories
interact, both inside and outside the firm. The stronger bonds among value
repositories will shape the firm’s core value creation architecture, which in turn is
linked by the weaker, second-order bonds, into the larger value system of the firm.
CHAPTER 3. BASIS FOR A NEW APPROACH
41
In a rather similar way as ecologists observe, we should expect that a firm
ecosystem had strong interactions among its key value repositories, and weak
interactions across its boundaries (Jørgensen 2006). For the purpose of this thesis
field research (Chapter 5), the intensity of the interactions between the airlines
value repositories are scaled in five different levels: Zero, Very Weak, Weak, Strong,
and Very Strong.
Type. Interactions between value repositories can be of two main types: intra-
component interactions, and inter-component interactions. Each type of interaction
usually features a distinctive level of intensity and dynamics. As Simon (1962)
observes, inter-component interactions are generally of a lower intensity than intra-
component interactions. Although this may be true in the field of natural sciences,
further analysis remains to be done in the case of the firm. In some cases, inter-
component interactions may be so weak that some value repositories might be
studied in a stand-alone fashion. For example, in the thesis field research (Chapter
5) the main type of interactions investigated are those among value repositories
from inside the firm, these being influenced by up to 10 environmental constraints.
Dynamics. Interactions among value repositories can be sorted out by high and low
frequency interactions. High frequency interactions usually take place inside value
repositories, whereas low frequency interactions are more common among
different value repositories.
Decomposing the interactions among the firm’s value repositories is an iterative process
that involves the particularization of both the strong and weak interactions (see Chapter 5). This
process may start focusing on the intra-component level of interactions and, as the process
CHAPTER 3. BASIS FOR A NEW APPROACH
42
unfolds, make refinements to address the next round of inter-component interactions. Note
that a parsimonious approach, as the 4-round Delphi process used in this thesis field research,
should be used for decomposing the interactions and strike a balance between excessive detail
and simplicity.
Eventually, the analysis and categorization of the interactions among the firm’s value
repositories should allow researchers and practitioners not only to characterize the interactions
but also to obtain a measure of the “Total Interaction Capacity (TIC)” of the firm as a whole —or
of a particular value repository. The TIC would be contingent on the three decomposition
variables described before —intensity, type and dynamics— and may be used to determine the
maximum number of simultaneous interactions that a value repository, or the firm as a whole,
may assume at any given time. Limitation of the firm to continue interacting may be a sign of
the firm’s own information processing capacity.
3.5.3 Modeling of the firm’s network dynamics
Once the firm’s hierarchy has been revealed and the interactions between value
repositories decomposed, the next step in the new theorizing process based on complexity
should focus on accomplishing a tentative modeling procedure. This way we may offer new
possibilities for the exploration of key behavioral questions, simulate specific firm’s behavioral
scenarios, and investigate new theories of the way the firm behaves.
Furthermore, modeling compels researchers to formulate hypotheses, determine what
data are available and what data are needed, and assess the degree of understanding about key
components and interactions of the firm. Altogether, modeling is a big step forward in the run
to get a better understanding of complexity of the firm.
CHAPTER 3. BASIS FOR A NEW APPROACH
43
Several model formulations can be envisaged —e.g. dynamic models, stochastic models,
agent-based models, fuzzy models, etc.— and the ability to choose among them requires that
sound constraints are imposed on the model not to make it more complex than the data can
bear; in other words, researchers must manage the trade-offs between knowing much about
little or little about much (Jørgensen, Bendoricchio 2001). In some cases, there might be
environments where the levels of complexity imposed cannot be managed by firms, unless they
set considerable simplifying constraints on the information processed.
To illustrate the above considerations, it may be useful to start working with tools such
as the adjacency or incidence matrix, a matrix made of ones and zeros, which assigns a value of
1 to the interactions between an element with another , and 0 to the absence of any
interaction. Most real-world incidence matrices are sparse and they usually concentrate the
larger interactions close to the diagonal of the matrix. Incidence matrices can also be useful to
infer subsystems with different degrees of clusterization (Simon 1976).
After formulating a model of the value system of the firm, researchers will need to
undertake verification tasks in order to conveniently assess the behavior of the model, as well as
to check how the model reacts before changes of inputs and/or of the basic assumptions.
Finally, validation of how well the model outputs fit the data should also be performed. As we
shall see in Chapter 4, these modeling activities are covered within Stage 4: Modeling and
Simulation of our proposed CBVF methodology.
THIS PAGE INTENTIONALLY LEFT BLANK.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
45
CHAPTER 4.
COMPLEXITY-BASED VIEW OF THE FIRM
4.1 Introduction
The CBVF’s subject matter is not different from that of many other theories of the firm:
the study of the behavioral phenomena produced by the firm as a living social and economic
structure.
If I may use the simile of systems biology, the CBVF approach aims at understanding
how the behavioral properties of the firm are brought about by the interactions of their
constituents (Kitano 2002, Alberghina, Westerhoff 2007). In other words, the CBVF aims to
decipher how the components of the firm (value repositories) jointly bring about firm behavior.
The problem addressed by the CBVF is therefore one of understanding the behavior of
the firm that does not rely on simple enumeration of its components and processes, thereby it is
perhaps necessary to define what we regard as constituting “understanding”.
For the purpose of obtaining an actual understanding of the behavior of the firm,
extracting and elucidating the general properties of the firm, together with a simplified,
condensed description that removes nonessential details through abstraction, is necessary
(Kaneko 2006). However, the CBVF is not an approach that simply reduces everything to a
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
46
description of the firm’s components. Instead it is an approach that seeks the properties of the
firm as a complex system, using a coarse-grained description of the properties emerging from
collective interactions between components.
In order to achieve that goal, the CBVF aims to envision systemic firm behavior from
their constituent value creation and value exchange processes , rather than by describing them
independently. The premise is that there is something to be inferred from the firm constituents
that cannot be discovered and understood by economics alone; properties that are embedded
in its components alone, but that emerge from its network collective behavior (Kaneko 2006,
Boogerd, Bruggeman et al. 2007).
Ultimately, the CBVF approach is concerned with the relationship and interaction
between value repositories, the firm as a whole, and its environment. It treats value repositories
as nodes in an organized network system with functional and behavioral properties. It uses
models to describe particular value repositories and their links to arrive at new explanations of
the firm. And last but not least, it is concerned with explaining and envisioning firm behavior on
the basis of the value repositories behavior.
Hence, a fundamental pillar of the CBVF is to seek patterns that can be used for
explanatory and predictive purposes —i.e. of the way the firm behaves, operates, organizes, and
evolves— and not scientific laws that explain the behavior of the firm in a deterministic way.
Consequently, precise and comprehensive experimental analyses of the firm, at levels between
the firm-system level and its value repositories, is a requirement for the CBVF, as it should also
be the accurate interpretation of the resulting experimental data.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
47
The above entails that the CBVF approach does not conform entirely to the “canonical”
―more naturalist― way of building up theories, as it is not expected to contain lawful
statements. Paraphrasing Morgenbesser, to say that the firm is complex may be a well
confirmed statement but a too general and commonplace one to be credited as a scientific
approach. Many times such lawful statements are simply not useful for the tasks that a good
theory of the firm needs to accomplish (Morgenbesser 1966).
Of course, this does not mean that our proposed CBVF does not aim to become a
genuine scientific approach to understanding the behavior of the firm. Quite the contrary, what
this author contends is the need for the CBVF to look beyond lawful statements, as they are not
of unique interest to the understanding of the firm. We want to know not just that the firm is
complex, but also the reasons that explain why the firm is so complex, the factors contributing
to firm’s complexity, and what complexity entails for the firm and its environment.
For all said above, the aim of this chapter is to tackle the problem of complexity of the
firm as a more comprehensive extension of the preceding theories of the firm covered
elsewhere in the literature. In this chapter we introduce the fundamental features constituting
our proposed CBVF, as an extended and complementary approach to the conventional theories
of the firm, yet with a deeper analytical scope towards the reality of the firm.
As we shall see, the CBVF is an approach to old and new problems, which builds upon
valuable ideas of the past, but also breaks even in the field of complexity with some tools and
ideas borrowed from more experienced scientific disciplines. Furthermore, in our struggle to
comprehend the behavior of the firm within its complexity, this chapter proposes a theoretical
characterization of the CBVF for the first time. On the basis of this characterization, a
methodological framework to tackle complexity from a practical angle is also presented.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
48
The ideas contained in this chapter should contribute to our overall understanding of
the firm as a complex system and, paraphrasing Einstein’s popular quote, to figure out how to
practically think about the complexity of the firm.
4.2 Theoretical Constituents of the CBVF
Complexity has remained somewhat invisible to most prominent theories of the firm, to
the extent that when firms’ theorists stumble upon complexity they usually use vagueness and
ambiguity, or they refer to issues that are simply “hard to explain with more traditional
economic methods” (Amman, Tesfatsion et al. 2006).
For this author, one of the major reasons explaining the absence of a comprehensive
approach to complexity within the theory of the firm is the lack of a framework of reference that
assists scholars and practitioners in properly characterizing complexity, in objectively featuring
its subject matter, and in providing practical guidance to cope with it. These are precisely the
goals that the CBVF aims to cater for.
Upon considering that every scientific discipline has its own particularities that
intrinsically define a complex system (Sanders 2003), and that the notion of complexity cannot
be featured universally and be the same for all disciplines, in this section we abstract some of
the key features that may characterize the firm under the CBVF approach. Such a list, inevitably
idiosyncratic from our perspective and not comprehensive in aspiration, contains only those
features that characterize the firm as a complex system and are foundational to the CBVF
approach:
Value as the basic building block of the firm.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
49
Nonlinear behavior.
Large number of interacting components.
Hierarchical structure.
Multi-level network dependency.
Descriptive-predictive uncertainty.
Structural homeostasis.
Differentiation by specialization.
Set forth below is a more detailed description of each of the constituents of the CBVF.
4.2.1 Value as the basic building block of the firm
Within the CBVF approach, we take the notion of “value” as the basic building block of
the firm. What this means is that “value” is the fundamental element from which the firm’s
interacting components (value repositories) originate, differentiate and become cohesive. In
other words, “value” is the brick and mortar of the firm’s architecture, without which the
components of the firm would inevitably collapse.
The discussion on the conceptualization of “value” lies at the very heart of economic
thought. Most likely no other single term in the history of economics has engaged the minds of
so many thinkers and philosophers as the theories surrounding the notion of “value”, the
concept thus becoming a recurrent topic in the most influential schools and streams of
economic knowledge.
But, what is the intrinsic reason why “value” is such a relevant concept? We might
answer this question arguing that understanding “value” reveals the foundational nature of
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
50
economic life and the very basic structure of economic facts. In fact, all theories of value known
to this author, from the oldest to the newest, aim at understanding and characterizing the
nature of value in economic systems, as well as focus on the whys and hows value is produced
and its localization within the economic system. Furthermore, for most theories what matters is
not really that “value” can be isolated in one way or another, but rather that “value” is the
outcome of a complicated web of interactions and relations between the various components of
an economic system.
In addition to the above epistemological arguments, the term “value” plays a central
role in the firm’s decision-making process, and is also a focal variable in the creation of
competitive advantage (Day 2002). In both cases, the “forces” behind value creation —i.e.
technologies of production, intersection of utility and cost, etc.— are considered key factors
that explain the distribution of income and economic growth. Last but not least, the relevance of
value is such that for many practitioners in the business arena, the maximization of “value” is
not only a vision or strategy, but rather the sole purpose of the firm and the scorecard for the
organization (Jensen 2002).
More evidence on the prominence of “value” as the basic building block of the firm is
provided by recent investigations, which demonstrate that customer-perceived value is a
cornerstone in relationship marketing and customer loyalty (DeSarbo, Jedidi et al. 2001, Day
2002, Priem 2007). Some authors even go as far as to compare value creation with the
customer's use value, or customer’s utility (Bowman, Ambrosini 2000); and yet for others,
customers are key value creators and they play an active role in co-producing value (Parolini
1996, Ramírez 1999).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
51
Consequently, if we are to tackle complexity of the firm we must first gather a good
understanding on the problematic of value for the customer (Woodall 2003), as a key driver of
the firm’s value dynamics and given its proven capacity to tie “the surface phenomena of
economic life to some inner structure or order” (Heilbroner 1983). As we shall see next in the
thesis, unraveling the multidimensional value system dynamics ―as described in Section 4.3.5,
B.1― using the analyst’s practical knowledge and experience becomes particularly important for
the later application of the CBVF approach.
4.2.2 Nonlinear behavior
Nonlinearity is a key feature that characterizes the behavior of the firm as a complex
system and, as an extension, the CBVF approach. A nonlinear behavior occurs whenever a
response is neither directly, nor inversely proportional to its cause. Themes invoked by this
notion include process, emergence, and ongoing, perpetual novelty (Meyer, Gaba et al. 2005).
Additionally, a system is nonlinear when it does not satisfy the superposition principle,
meaning that the output of a nonlinear system is not directly proportional to the input. This, in
turn, involves that the system does not satisfy the homogeneity and additive properties
together —inversely, a linear system is one that satisfies both properties.
The property of homogeneity defines a homogeneous function of degree λ as a function
such that for all points in its domain and all real , the equation
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
52
holds, where λ is a real number. It is assumed that for every point in the
domain of , the point also belongs to this domain for any (Kudryavtsev
2001). As we can see this is a function with multiplicative scaling behavior.
The additive property means the system preserves the addition operation for any two
elements x and y in the system, thus:
If input produces and input produces , then input produces .
Similarly, a nonlinear system of equations is one in which the equation(s) to be solved
cannot be written as a linear combination of the unknown variables or functions that appear in
the system. The differential equations governing some systems, including some thermal, fluidic,
or biological systems, are nonlinear in nature.
Nonlinear science has its origins in Poincaré’s solution to the "N-body Problem" nearly a
century ago —the problem of predicting the individual motions of a group of celestial objects
interacting with each other gravitationally. However, it is only recently that nonlinear
explorations have become commonplace in a widespread number of scientific disciplines.
Gradually scholars have begun to suggest that nonlinear processes are much more ubiquitous
than we could have ever imagined, and that they are a fundamental feature of natural systems
(Daneke 1997).
Nonlinear dynamics is characterized by emergence, self-organization and evolution, all
of them concepts familiar to our proposed CBVF. Emergence refers to new properties that were
not present in, or predictable from, the initial conditions (Holland 2000, Stace, Goldstein 2006).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
53
Emerging processes arise from the interaction between components of the system, which make
it impossible to predict future states. Emergent conditions allow the system to self-organize and
acquire a new order, thus making self-organization a mechanism for systems evolution
(Bohórquez, Espinosa 2015).
What this author suggests is that there is compelling evidence by which the firm would
behave in a nonlinear manner. Factors such as the use of advanced information technologies,
the organization of resources in complex networks, or the development of innovation activities
that affect productivity and transaction costs, to name just a few, provide evidence as to why
the superposition property would not be met by the firm.
Notwithstanding, it is worth noting that being subject to nonlinear dynamics is not a
necessary condition for the firm to be considered a complex system (Ladyman, Lambert et al.
2013). Actually, we can find examples of network systems performing linear behavior, which are
studied by complexity disciplines. Moreover, it is also perfectly possible to think in a linear way
about systems which exhibit nonlinear dynamics —i.e. complex systems subject to game-
theoretic, or quantum dynamics subject to linear dynamics (McKay 2008).
4.2.3 Large number of interacting components
“Many more than a handful of individual elements need to interact in order to generate
complex systems” (Ladyman, Lambert et al. 2013). What this assertion highlights is that
complexity emerges when a large number of components are present in a system and,
subsequently, if they are engaged in many interactions. Upon relying on the above premise, we
might then wonder, what is a “component”? and, why does this author submit that the firm has
a large number of them?
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
54
Most definitions of complexity from various scientific disciplines establish the large
number of components as a key property of complex systems. However, when authors
approach the idea of “component” much of vagueness and ambiguity is found. In fact, both
terms ―complexity and component― invoke problems that are not unique to economics or the
theory of the firm, but which are also characteristic of other highly respected scientific
disciplines, such as systems biology or quantum mechanics.
In the complexity jargon, a “component” is usually assimilated to the fundamental
functional unit of a system and, as such, it may be built up and broken down. Relations or
“forces” among components make them interact either in a static or dynamic manner, and they
can adopt a heterogeneous configuration as well.
Under our proposed CBVF, the firm’s components are represented as “value
repositories” (see Section 3.5.1). Value repositories group together those activities, processes
and resources aimed at creating unique exchangeable value, thus exhibiting a duality in their
behavior between value creation and exchange (Section 3.5.2). It is worth noting that it is the
very existence of value in the firm-system, and the flows of exchange, what matters most to the
CBVF, and not the particular incentives ―i.e. maximization, optimization― or the managerial
actions needed to create and exchange value ―the latter falling out of the scope of this thesis.
Notwithstanding the foregoing, value repositories are not constructs that can be
unequivocally defined by any observer from inside or outside the firm, but components that
reflect a real firm’s value system dynamics, its structural and behavioral attributes being
dependent on the particular rationale of value made by the firm.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
55
Furthermore, uniqueness is a key principle that researchers and practitioners should
carefully observe when they attempt to symbolically represent value repositories. This basically
means that no two different value repositories should create and exchange the same value, thus
reflecting the intrinsic division and specialization that exist within the firm’s value system
dynamics.
Therefore, as far as we make sure that no two value repositories cover the same value
dynamics (Fig.6), we will be able to avoid redundant components analysis and creating a
misleading or overly too complex picture of the firm’s value system dynamics. For example, two
value repositories previously outlined as delivering up-to-date performance information on
different products to a customer should be symbolically represented as a single value repository
under the CBVF analysis, as their value dynamic is the same.
It is also important to note that value repositories take the form of nodes in a complex
web of networks in continuous interaction with one another. The composition, boundaries, and
processes comprising the firm’s value repositories (VRs) are determined by the idiosyncratic
(value creation and exchange) interactions that are characteristic of every firm, and the
particular path taken in creating and capturing value. Without the logic of a network of
interacting VRs, the firm would merely become a meeting of independent individuals with no
feasible goals to attain.
Ultimately, the firm is made up and interacts with a large number of interacting VRs, not
only from inside the organization, but also from its outside environment. Moreover, not all the
VRs are intrinsically of the same kind ―i.e. some may be mainly focused on internal customers,
whereas others may be more focused on external customers― nor do they interact in the same
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
56
way ―i.e. some may pursue exchanges for money while others may prioritize exchanges of
information or other specialized resources.
As we shall note, the complex network of interactions among VRs thus formed increases
the difficulty to grasp the behavior of the firm as a whole and hinders any attempt to build any
theory of the firm in a simplistic way. At the end, containing a large number of interacting VRs is
a necessary but not sufficient condition for the firm to be deemed as a complex system.
4.2.4 Hierarchical structure
Firm’s VRs form a structure of hierarchical levels. This means that the firm is structured
in a variety of levels that interact with the levels above and below, and may exhibit some kind of
causal regularities and predictable behavior (Ladyman, Lambert et al. 2013). Hierarchical
structures similar as that of the firm, abound in living and non-living systems, from natural
ecosystems to stars and galaxies.
Nonetheless, the VR-based hierarchy of the firm does not convey a structure of power
(Simon 1962b, Simon 1999), but rather a network-within-network arrangement of interactions
among tangible (people, resources) and intangible (rules, processes) factors, all of them
attracted by the same “guiding force”: the creation and exchange of value. Within this hierarchy,
each level is connected with another by feedback mechanisms, thus forming one dissipate
structure where the cycle moves from the lower to the higher VRs and back (Karsten 1990).
Unlike in the hierarchical structures of physical and biological systems, the higher levels
in the hierarchy of the firm may not necessarily assemble all levels below under special
circumstances. This does not mean that the “downward causation” or “emergence” property
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
57
characterizing natural complex systems is no longer valid under the CBVF. Instead what this
author suggests is that the resulting dynamics of the firm’s value system can be very complex
and unpredictable, therefore not always the value exchanged in the higher levels is the sum
total of the value created in the levels below. For example, some value created at the levels
below may “dissipate” in non-productive processes when scaling up to the higher levels of the
firm, or the managers at the higher levels may decide not to fully incorporate all value created in
the lower levels into their Value Offerings, based on their particular vision of the market
conditions, or tactics with specific segments of customers.
Hierarchical processes also help explain why and how the firm evolves from a grouping
of individuals to a complex system. As Simon points out, in the process of evolution each
intermediate level forms a stable configuration, which may be selected for further levels to build
on top (Simon 1999). In this way, the hierarchy of the firm may well be explained by the
efficiency and stability of a hierarchical building process (Ladyman, Lambert et al. 2013), where
higher levels are built on top of lower levels following a gradient of value exchange.
From a practical perspective, the idiosyncratic hierarchical structure of the firm can help
generate meaningful and predictive theories for higher levels without knowing much about the
lower levels. This is particularly the case in those firms where a particular higher level of the
hierarchy can be explained with only a broad picture of the VRs below or with no picture of
them at all.
Notwithstanding, in more complex firms —i.e. transnational corporations— a good
understanding of the particular levels above and below is critical to grasp a comprehensive and
idiosyncratic view of complexity of the firm. This is so because the value dynamics phenomena
in highly complex firms are well beyond the scope of any particular level of the hierarchy.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
58
4.2.5 Multi-level network dependency
As described above, complex firms contain a large number of VRs structured in
hierarchical levels in continuous interaction. From a CBVF perspective, VRs are represented by a
number of nodes (vertices) and its connected relations (edges), the later corresponding to the
flows of value exchange. Furthermore, firm’s VRs are themselves nodes in higher levels of the
hierarchy, forming a “web of networks” which hosts the firm’s value system dynamics (Fig.2).
This conceptualization of the firm transcends conventional theories of the firm, which
mostly do not offer a glimpse on networks, and somewhat resembles Thorelli’s own idea of the
economy as a network of organizations with a vast hierarchy of subordinate, crisscrossing
networks (Thorelli 1986).
The multi-level network structure in which the firm becomes, entails that the VRs can
only be understood if we account for all the internal and external VRs to which each VR is
recursively linked, eventually forming the firm as a “whole” (Mella 2009). In other words, each
VR takes on significance in the context of the relationships with the sub-networks it is made up
of, and the supra-network higher level structure to which it belongs to.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
59
Figure 2. Illustration of the Multi-level Web of VRs
Source: own elaboration
This multi-level network dependency also involves that any change in a VR will be
accompanied by changes in the entire network structure, and in all the superordinate and/or
subordinate levels (Mella 2009). This in turn may invoke the principle of “co-evolution”, as the
individual part (the VR) and the whole (the firm) exist by unfolding one another (Karsten 1990,
North 1990, Nelson, Sampat 2001). Co-evolution ultimately implies that the development of VRs
mirrors the own evolution of the firm, and vice versa. That is, VRs and the firm evolve together
as a whole, with VRs prompting changes in the firm —and vice versa— as a function of time.
Multi-level network dependency is also related to Simon’s “near-decomposability”
property of complex systems. According to Simon (1999): “a much higher frequency and
intensity of interaction takes place between components belonging to a single sub-system than
between components belonging to different sub-systems; and this principle holds for all levels of
the hierarchy”. From Simon’s proposition we may infer that if a system is disturbed, then all
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
60
subsystems at the lowest level will come to a steady state before the subsystems at the level
above. Consequently, the whole system could be described in terms of the average behavior of
the subsystems, and more particularly by their principal eigenvalues.
Furthermore, a nearly decomposable system should allow us to factor the system, which
in turn would imply not having to deal with all of the system complexity at once. The above has
important practical implications, because as Simon notes: “having determined the behavior of
subunits at one level, we can replace the details of these subunits by a small number of
aggregate parameters, and use these to represent the system at the next level above”.
It is likely that Simon’s assumptions did not fully considered the networked structure of
complex systems, but instead one of boxes-within-boxes, which makes a significant difference
when we try to apply it to our CBVF approach. Networks differ greatly from boxes, insofar as
complex firms do not seem to follow the quasi-independent behavior attributed by Simon to
network components. Remarkably enough, VRs are mutually interconnected with other VRs,
from which they depend and obtain continuous feedback.
In this context, what it seems clear is that any attempt to understand the behavior of
the firm from a network perspective requires a shift in focus away from the way the firm
allocates and structures its internal resources, and towards the way it relates its activities and
resources to the other parties constituting the network.
What is more, once a network view of the firm is adopted, considerable changes occur
with respect to the basic assumptions made by the conventional theories of the firm, among
which it is worth mentioning the own definition of the boundaries of the firm, the assessment of
the firm effectiveness, or how we manage the firm (Hakansson, Snehota 2006). Ultimately, a
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
61
reconsideration of the very essence of the notion of firm would be necessary, although we will
leave this discussion aside.
4.2.6 Descriptive-predictive uncertainty
Investigating into the behavior of the firm as a whole, and into the relations between
the various VRs, always involves the observer (or researcher) in an essential way. The human
observer constitutes the final link in the chain of the observational process, and the complex
properties of the firm can only be understood in terms of the interaction between the observer
and the firm itself.
This premise, inferred from Heisenberg’s uncertainty principle (Heisenberg 1958), would
assert the impossibility to describe and, therefore, to predict with entire certainty, how a firm or
a particular VR will behave at a particular moment of time and how a particular value dynamic
will come to occur. All we can do is to describe the conditions under which a concrete behavior
of the firm will most probably come to happen.
The descriptive-predictive uncertainty property would also explain that we cannot know
exactly all the properties of the firm, since whenever one property is determined precisely, the
other properties will become uncertain or will need to be set as ceteris paribus —note that the
ceteris paribus assumption is a reductionism approach that does not work satisfactorily within
the CBVF. In other words, as in the case of complex physics systems, the uncertainty property
acts as a limit on the exact and whole knowledge that we can have on the behavior of the firm
(Karsten 1990).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
62
This apparent indescribability/unpredictability that characterizes the behavior of the
firm may be due either to hidden variables, to our ignorance in understanding the factors
accounting for a given structure of value repositories, or to the difficulties to recognize the
interrelationships between the VRs dynamics. Consequently, the firm cannot longer be viewed
as a deterministic whole system, but instead only be apprehended through complex, most
probable, and sometimes paradoxical views.
Furthermore, under our proposed CBVF approach, the firm cannot be disaggregated
into simple, isolated, and independent components (VRs) with no relation among them and its
environment. As far as the CBVF is concerned, everything in the firm becomes an integrated and
an interconnected whole.
4.2.7 Structural homeostasis
The notion of homeostasis is a much debated one in the field of complex systems in a
broad variety of scientific disciplines. First originated in physiology, homeostasis was later
adapted to the social and economics fields. Hou-Shun definition of homeostasis considers it as
the constant act of balancing of two opposing forces to maintain stability in the most developed
organism, be it biological or economic (Hou-Shun 1956). Defined in this manner, homeostasis
goes well beyond closed (isolated) systems equilibrium to imply open systems dynamics through
the description of various self-adjustments (Bailey 1990).
Other authors, such as Simon, define homeostasis as the property of a complex system
in which “by means of feedback mechanisms or by other methods (…), a system may be able to
hold the values of some of its important properties within narrow limits, and thereby greatly
simplify internal processes that are sensitive to these properties” (Simon 1999). More recently,
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
63
other definitions of homeostasis, as the one made by Paradice when describing modern
exchange-based societies, denote how societies are able to maintain a relative overall constancy
despite complex internal changes (Paradice 2009).
Specifically, our CBVF approach uses the term homeostasis to describe the process by
which the firm is able avoid substantial changes of its internal basic structure, despite significant
changes in the external environment. According to this definition, both the variability and the
unpredictability of the environment impose a need for homeostasis, otherwise the firm would
be doomed to vanish every time serious environmental changes take place. In other words, we
might say that homeostasis enables the firm to hold up its key hierarchical structure and value
system dynamics, despite changes in its environment, thus ensuring its operational continuity.
Although homeostasis is an inherent (an inherited) property of complex natural systems,
this is not always the case in the firm. Homeostasis necessarily needs to be instilled into the
design of the firm to yield the expected outcomes. This generally involves developing some kind
of “layer” that greatly attenuates the transmission of environmental changes into the interior of
the firm. The same principle might apply to the way the firm shields the various VRs from each
other.
When properly realized, structural homeostasis is a method of reducing the firm’s
complexity, though at the cost of some new complexities that may appear in the form of the
homeostatic mechanisms themselves. Such mechanisms would include the development of self-
learning and auto-adaptive processes within the firm, as well as the transformation of the
boundaries of the firm into specialized interfaces.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
64
Note that the presence of structural homeostasis in the characterization of the firm is a
convenient but not sufficient condition for the firm to be considered a complex system.
4.2.8 Differentiation by specialization
As seen in the previous sections, VRs are, in essence, specialized subnetworks within the
larger firm-system network for the creation and exchange of value. Although at first the VRs
configuration of the firm may seem to contribute to simplification of the behavior of the firm,
they add new complexities in the form of specialized mechanisms, coordination mechanisms,
and exchange interfaces.
Specialization responds to the need of the firm to accomplish very specific value-driven
tasks in order to operate successfully and with increased productivity in its environment. In fact,
if we assimilate VRs with agents of a social system, we might notice how they become more
specialized as the group (firm) size increases and there is too much task choice available
(Cockburn, Kobti 2009).
According to Cockburn and Kobti, given a system with a large number of tasks and a
small level of connectivity, agents are much less likely to choose tasks not being performed by
others. Hence the need to combine a set of VRs highly connected, each capable of performing
one —or a few— of the firm’s value-related functions, and connect them so that they can
cooperate with each other.
The number of VRs and the nature of the interactions among VRs are an approximate
measure of complexity of the firm, given that the firms under comparison share the same notion
of value and dynamic of interactions. Less complex firms would show just a few VRs and a low
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
65
number of interactions among them, whereas as the number of VRs increases together with the
number of interactions, the complexity of the firm becomes higher.
It should be noted that VRs specialization is mainly a matter of design within the firm.
Simon (1999) even contends that “specialization should be carried out in such a way as to keep
the interactions between the specialized components at as low a level as possible”. This certainly
should help keep the complexity of the firm within a manageable level. Nonetheless, Simon’s
recommendation weakens whenever we consider the firm’s VRs subject not only to internal
design, but also to external factors facilitating or blocking the value dynamics.
As a consequence of the above, even for basic value exchanges, the design of the VRs
specialization becomes a rather complex task that requires choices to be made at each level of
the hierarchy of the firm. These choices would include identifying which dimensions of
specialization are more important, and under which VRs certain tasks should be performed
together.
Other elements affecting the specialization of VRs are its boundaries. In fact,
understanding VRs boundaries has important practical implications and is key for the
characterization of the firm as a complex system, even though they are more blurred and
difficult to trace than ever (see Chapter 3). Note that we refer to boundaries in plural, thus
presupposing that under the CBVF approach the firm does not feature a single, uniform, all-
purpose, high level boundary, but as many boundaries as VRs exist, each specialized in the
creation and exchange of a particular type of value.
The importance of the VRs boundaries resides in that they act as specialized agents for
exchange of value in and out of VRs, thus affecting the firm’s configuration of its own network
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
66
context. Boundaries differentiation and specialization is essential for an efficient and effective
value dynamics. Moreover, specialized boundaries simplify the process of interaction between
VRs, as the need to constantly adjust the interaction interface at a VR level is significantly
reduced.
Specialization, and the resulting differentiation of VRs, is thus a convenient but not
sufficient condition for the firm to be considered a complex system.
4.3 Methodological Framework
The growing interest in the complexity of (living and non-living) systems has proved to
be useful in many non-economic disciplines, facilitating some remarkable advances in scientific
knowledge and further developing our ability to get through disciplinary boundaries. However,
the term “complexity” has entered not only a stage of overuse, but its meaning and utility has
started to dilute even before realizing its potential. Some authors even assert that complexity is
falling a victim of its own success (Crutchfield 2008).
The times where it was enough to say that this or that was “complex” with hardly any
practical consequences for the firm, are now gone. Going one step further has become
inevitable, all the more necessary. A step ahead must be taken with the chief intention of
operationalizing the notion of complexity in the firm and provide the means to inform
professional practice.
In this spirit, this section specifically provides a methodological framework to bridge the
gap between the conceptual use of our CBVF and its instrumental adoption. Furthermore, the
method of the CBVF shall be concerned with how the firm may be designed to be as simple and
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
67
efficient as possible —in structure and processes— as to survive in a complex environment,
where it interacts continuously in competition with other firms performing a wide range of
adaptive functions (Simon 1999).
Accordingly, the methodological framework here presented strives to provide a reliable
basis for the systematic analysis of complexity of the firm, and approximate decisions that might
be converted into actions. In terms of Gibbon’s et al. modes of knowledge production, the
practical-oriented solutions provided by our CBVF methodological framework aim at combining
the scientific rigour normally associated with Mode 1 knowledge, and the problem solving
orientations of Mode 2 (Gibbons, Limoges et al. 1994).
4.3.1 Design
The design of a methodological framework is a crucial step on the path towards
incorporating complexity into the investigations of the behavior of the firm. As an essential part
of the CBVF general approach, our proposed methodology has been thought to enable
researchers and practitioners to tackle complexity from a theoretical-practical perspective, and
get a more realistic insight into the behavior of the firm than that provided by other
conventional theories of the firm.
The CBVF methodology represents a framework within which particular activities,
methods and tools are to be deployed, all of which have been selected among a broad range of
different options available in the field of complexity studies.
Originally formulated in response to the need to overcome the serious utilization
problem of academic management theory (van Aken 2004) and to tackle the problem of
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
68
complexity of the firm, the rationale of the CBVF methodology resembles the principles of
design science (van Aken 2004), pragmatic science (Anderson, Herriot et al. 2001), and action
research (Järvinen 2005). Consequently, the methodology focuses on the process to understand,
design and experiment actions to be later realized within the scope of the firm, and it is not too
much interested in “what is” but in “what can be”.
Drawing upon the methodological underpinnings referred above, the resulting product
of the CBVF methodology has the character of a “prescriptive scenario” or, in Bunge’s words,
“an instruction to perform a finite number of acts in a given order and with a given aim” (Bunge
1967), rather than a formal causal model where one or more dependent variables are strictly
explained in terms of one or more independent variables.
The assembly of a “prescriptive scenario” that can be set out by researchers and/or
implemented effectively by practitioners of the firm is thus the raison d'être of the CBVF
methodology, which purposefully requires the execution of the following four designs in an
iterative way:
1. The firm design, or the properties underlying the structure and dynamics of the
firm-system and the logic for which we need to know the properties and settings
that support the actions to be deployed. This is chiefly achieved by mapping the
theoretical constituents that characterize the firm as a complex system (Section
4.2).
2. The scope design, or the breadth of the actions to be deployed based on the
knowledge of the level of complexity of the firm and of the interrelations among the
firm’s value repositories. This is mainly accomplished by visualizing the complexity
of the firm and, specifically, by characterizing the network representing the firm.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
69
3. The process design, or the mechanisms to be used for deploying concrete actions in
the real world. This is accomplished by modeling and simulating alternative
prescriptive scenarios in a multiple case cycle. By testing artificial scenarios
researchers and practitioners might gather further insight into the indications and
contra-indications for the application of a particular set of actions (van Aken 2004)
and, therefore, decide between different deployment strategies.
4. The improvement design, or the re-engineering of the firm’s structural, relational
and/or dynamical components. This is accomplished by optimizing the firm’s
operating logic, once the effectiveness of the actions deployed has been simulated
in the context of its intended use, and the desired results have been adequately
assessed.
As stated above, the main goal of the CBVF methodology is to come up with a
“prescriptive scenario” formed through an iterative experimentation process, which develops
knowledge to be eventually implemented and refined in successive phases by researchers and
practitioners of the firm.
Ultimately, it is worth noting that one of the things that distinguishes the CBVF
methodology from other conventional methods is, besides its flexibility, that we need the
implication of practitioners. That is, if the CBVF methodology is to bridge the gap between the
theory and practice of the firm, as well as translate the study of complexity into practical
insights for managerial practice, then research must result from the involvement with the firm
and the intention to take action over the basis of specific prescriptive scenarios (van Aken 2004).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
70
4.3.2 Hard vs Soft Approach
In general, complexity scientists have used either a “hard” or “soft” methodological
approach (Richardson, Cilliers. 2001). The hard approach assumes that reality is determined
and, hence, determinable using analytical science. One of the preferred means of investigation
within the hard approach is computer simulation, in which complex systems are modeled in an
attempt to uncover the conditions that underlie their emergence and to make sense of the new
capacities that arise once emergent (Davis, Sumara 2014).
Under this approach, researchers enunciate propositions that are formulated in the
language of sophisticated mathematical functions and differential equations, and make every
effort to encompass all the variables deemed as relevant to explain the behavior of a complex
system as a whole. The hard approach thus tackles complexity heads-on and its main goal is to
model the behavior of the firm in a mathematical way by identifying, setting up the
relationships, and quantifying as many variables as the researcher’s knowledge of the system
allows.
As pervasive as the “hard” approach has been in the history of economic theory, some
authors are very critical today of the relevance achieved by the results and the progress made
by this approach in crossing the chasm between complexity, reality and the firm (Von Neumann,
Morgenstern 1947, Lambertini 2013).
Meanwhile, the soft approach draws on the metaphors and principles of hard science to
interpret a system rather than to represent a reality (Richardson, Cilliers. 2001). Instead of
struggling with complexity, it seeks to avoid complexity by decomposing the complex system
into more simple components, provided there is no appreciable relationship among them. The
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
71
“soft” approach therefore aims to address the behavior of each component individually, using
either quantitative or qualitative techniques, or both.
Furthermore, the “soft” approach, as inspired by Simon’s reducibility principle, admits
that we are quite unable to perform the computations that should help us describe the system
as a whole without the aid of some kind of simplification by aggregation at a higher level (Simon
1999). Therefore, whilst the hard approach attempts to integrate as much of complexity as a
quantitative model allows; “soft” tactics involves removing as much of complexity as feasible
from the system.
The CBVF methodology presupposes a few of the key assumptions made by the soft
approach, hence it supports the hypothesis that complexity can only be “defeated” if we are
able to modulate the system in a coherent and meaningful way (Ethiraj, Levinthal 2004). In
other words, the CBVF methodology assumes that it is always more practical and, generally,
delivers better results, if we are able to decompose the firm-system into a (intermediate) level
of modularity, which balances between over simplification and excessive breadth that might
blind our capacity to comprehend important interactions between system components.
Consequently, the CBVF methodology is mainly a method of descriptive nature which
aims at gathering as much knowledge as possible about concrete “burning” questions
surrounding the behavior of the firm. Note that in approaching the complexity of the firm there
is no reason to assume the existence of shortcuts, and being too impatient to answer the more
general behavioral questions of the firm, or approaching the complexity of the firm as a whole,
would merely delay progress (Jaynes 1991).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
72
Before going into detail it should be noted that the method of the CBVF is not
presumably for every firm. In order to apply the method, firms should exhibit specific properties
—those outlined in Section 4.2— or demonstrate specific mechanisms for their development.
Furthermore, this author holds reasonable doubts that, at the moment of writing this thesis, the
CBVF methodology is appropriate for use in groups of firms, thus noting that further research
would be needed before we can assert that the same properties characterizing complexity in a
single firm would continue to apply to groups of firms.
4.3.3 Method
Now that we have outlined the main features associated with the design of the CBVF
methodology, in this section we sketch a tentative method that might help researchers and
practitioners to systematically disentangle complexity and contribute to a more realistic
understanding of the behavior of a firm.
Our proposed CBVF methodology is composed of four stages (Fig.3), which outline a
framework of different but interrelated tasks where each stage builds upon the outcome of the
previous one. The stages should be performed in a sequential order to get as thorough as
possible understanding of complexity in the context of the firm. Failure to follow one stage at a
time, or the suggested sequence, might result in misconception of complexity and/or a
misunderstanding of the behavior of the firm, leading to inaccurate practical consequences.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
73
Figure 3. Stages of the CBVF Methodology
Source: own elaboration
It should also be noted that the CBVF methodology is a tentative, elementary, non-all-
encompassing set of tasks to help researchers and practitioners tackle complexity of a firm. One
of the key advantages being that it is a systematic approach that cater for the most important
analytical factors. Needless to say that further research in the methodological field would be
required to provide further detail to the stages, or even add new ones.
4.3.4 Data capture
Obtaining meaningful and reliable data to attain the objectives, accomplish the
activities, and implement the methods and tools described in the Stages of the CBVF
methodology, is a real challenge that every researcher and practitioner will face sooner or later.
No matter how robust the firm’s actual data are, or how well data support the firm’s decision-
making processes, most probably the analyst will need to invest time in obtaining the right data
to feed the CBVF methodology, and reworking internal numbers ―if not to build a new data
system.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
74
The CBVF methodology is not easily applicable “as is”. Its implementation requires the
availability of extensive (and complex) value-related data, which in turn is likely to be minimal or
non-existent in most conventional firms. What further complicates the problem is that, quite
often, the dimensions of the actual data are incompatible, and data systems are designed to
serve different requirements than those required by the CBVF approach.
The first obstacle emerges when the firm is not organized in VRs and, consequently, the
firm’s data systems do not recognize value (and VRs) as a dimension for data collection. In other
words, given that conventional data systems mostly accumulate financial data and costs around
products and organizational units (Garrison, Noreen et al. 2006), there is no general method
thought to map accounting data onto VRs. The resulting lack of correspondence is thus a source
of difficulty for the analysts’ attempting to implement the CBVF approach.
Furthermore, the vast majority of real-life firms rarely collect any data on value flows
and, when they do, they often use rudimentary tools for modeling interdependencies ―i.e.
through allocations for services and transfer prices for components or products (Hergert, Morris
1989). Ultimately, if the firm is not structured around VRs much less likely is that it has an in-
depth knowledge of its own value system dynamics or assesses value interactions between VRs.
The problems with data do not disappear when we focus our attention on the budget or
balance scorecard (BSC), sometimes used as proxies for value-related data. Generally, the
budget does not measure value creation activities, nor is it thought to generate any value-
related data. Hergert and Morris (1989) suggest two main reasons for this: the first concerns the
categories of expense charged to the budget, most often mandated by external reporting
authorities; the second, concerns the items of expense included in each category. Therefore,
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
75
using the cost of an activity as a proxy of value is not trivial and would require substantial
reworking of the numbers.
Additionally, the BSC is not particularly helpful either. No matter how sophisticated it is
or the refinements involved, the BSC focuses on the strategic implications of performance
measures (Barber 2008). The original dimensions enunciated by Kaplan and Norton —financial,
customer, internal process, and learning and growth (Kaplan, Norton 1996)— and its later
developments in the form of strategic maps (Kaplan, Norton 2004), still remain insufficient for
measuring value in the firm.
So, how can we solve the problem with data as far as the CBVF is concerned?
If neither the most frequently models used for organizational design, nor the actual
structures of the firm, are consciously designed around a network of VRs ―as suggested by our
CBVF approach― then the existing firms’ data systems are unlikely to provide useful data for its
implementation. Actually, there is little more that can be done to solve the way in which value is
measured in the firm’s conventional data systems.
Upon assuming this situation, the next question that we might ask ourselves is, how can
we generate the right data needed to effectively feed the CBVF methodology? To answer this
crucial question, the author suggests to develop a value-driven data system, the implementation
of which might be based on three subsequent phases (Fig.4):
1. Phase 1: the firm would need to develop and share a common understanding of
what its key value statements and value drivers are; which of its activities, resources
and assets are responsible for creating value; how and who creates value in the
firm, and how, why and with whom the firm exchanges value. The firm should them
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
76
examine the constituents of customer value and accumulate quantitative and
qualitative data —from inside and outside the firm— around these value drivers.
2. Phase 2 (optional): Involves reworking the firm’s available accounting data to make
it usable by the CBVF methodology and its proposed tools and techniques. As there
is no universal method for “transforming” cost and accounting data onto value data,
the firm would need to internalize the conclusions gained in the previous phase and
act upon its findings. Some authors have provided frameworks based on different
measures and perspectives that illustrate how to perform this data reworking
process (Partridge, Perren 1994, Lieberman, Balasubramanian 2007, Biem, Caswell
2008, Haile, Altmann 2013). However, it is just recently that these methods receive
more attention and begin to provide new potential foci for value assessment and
measurement.
3. Phase 3: This phase is aimed at the longer term, as it builds upon the insights and
the lessons learned from phases 1 and 2. Here the firm gets focused on the
development of an entirely new value-driven data system to effectively support the
implementation of the CBVF methodology. The new information system, which
could be called “value scorecard”, could serve a two-fold purpose: i) it would
prevent the firm’s current cost and management accounting data from being
contaminated by value analysis requirements and, ii) it would help to cover the
CBVF’s data needs without recruiting an army of analysts to continuously rework the
accounting data. As shown in Fig.4, it is advisable that phases 2 and 3 run in parallel
for some time, prior to making the “value scorecard” available as the firm’s primary
source of data. All in all, the “value scorecard” would rely on the assumption that
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
77
instead of trying to use a universal accounting system for all purposes, the
implementation of the CBVF approach requires a data system specifically designed
to facilitate a robust value dynamics analysis on the basis of both qualitative and
quantitative measures (Town, Kyrillidou 2013).
Figure 4. Phased Approach to Resolve the Problem With Data
Source: own elaboration
In the interim of deciding what approach to use and how to build up the resources
needed to improve the firm’s data system, a relatively rapid and plausible method to obtaining
useful data on the firm’s value system and start working in the implementation of the CBVF is
expert knowledge captured through a Delphi method.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
78
The Delphi method is a versatile research method typically used for futures research in
areas where knowledge is incomplete. In fact, it has been satisfactorily used in a wide variety of
research areas as a method to develop, identify, forecast and validate data (Skulmoski, Hartman
et al. 2007).
The Delphi method involves an iterative survey of experts, where each participant
completes a questionnaire and is then given feedback on the whole set of responses. With this
information in hand, each participant then fills in the questionnaire again, this time providing
explanations for any views (s)he holds that were significantly divergent from the viewpoints of
the others participants (Slocum 2005). The explanations serve as useful intelligence for others.
In addition, (s)he may change his/her opinion, based upon his/her evaluation of new
information provided by other participants. This process is repeated as many times as is thought
to be useful. The idea is that the entire Experts’ Panel can weigh dissenting views that are based
on privileged or rare information.
In general, the Delphi method was invented to overcome some of the following
difficulties when carrying out research tasks such as ours (Linstone, Turoff 2002, Slocum 2005,
Balasubramanian, Agarwal 2013):
When there is incomplete knowledge about phenomena and both a quantitative
and qualitative research is desired.
When a problem does not lend itself to precise analytical techniques but can benefit
from subjective judgments on a collective basis.
When experts needed to contribute to the examination of a broad or complex
problem have no history of adequate communication and may represent diverse
backgrounds with respect to experience or expertise.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
79
When more experts are needed than can effectively interact in a face-to-face
exchange.
When time and cost make frequent experts group meetings infeasible.
When disagreements among experts are so severe or politically unpalatable that the
communication process must be refereed and/or anonymity assured.
When heterogeneity of the experts must be preserved to assure validity of the
results, i.e. avoidance of domination by quantity or by strength of personality.
Researchers and practitioners should note that no two Delphi studies are the same. As a
matter of fact there are many varieties of Delphi, ranging from qualitative to quantitative, as
well as mixed methods. Nonetheless, common to all varieties are some design considerations
that include deciding on the sample composition, sample size, methodological orientation
(qualitative and/or quantitative), the number of rounds, and the mode of interaction.
Considering all these choices should add rigor to the method, which in turn shall contribute to a
successful Delphi process and, by extension, to a deeper understanding of the firm’s value
system.
4.3.5 Stage 1: Mapping the Firm-System
As we have seen before in this thesis, most firms in the real world act and behave as
open systems embedded in complex environmental settings with which value is exchanged . By
adopting this view, firms acknowledge they share a set of characteristics that are definitory of
complex adaptive systems (see Section 4.2).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
80
A thorough understanding of the characteristics featuring the firm as a complex system
and the extent to which they become apparent in the firm, is thus of paramount importance on
starting the path towards incorporating complexity into the practical investigation of the firm.
Mapping the firm-system is the first stage of the CBVF methodological framework and,
as such, it strives to provide a characterization of the firm’s complexity defining characteristics
and how they operate in a particular firm. Ultimately, a systematic mapping of how this
theoretical constituents of the CBVF operate and of the firm’s value system dynamics should
pave the ground for the next stages in the CBVF methodology and for a successful application of
the CBVF approach.
A. Objective and scope
Mapping the firm-system fundamentally involves outlining the firm’s complexity
constituents in a real context. The objective being to obtain a thorough understanding of the
role played and the interactions shown by each feature examined in Section 4.2, both
individually and in relation to one another.
In particular, given that the firm may be seen as an interface between an “inner” and an
“outer” environment (Simon 1996), researchers and practitioners should focus on the relation
among three key elements:
the purpose or goal of the firm.
the components (value repositories), and
the environment in which the firm operates.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
81
The purpose or goal shapes the firm’s inner environment, which itself is organized as a
network of internal value repositories (VRs); namely, a network of nodes that group together
activities which create unique and exchangeable value (Section 3.5.1). According to our CBVF
approach, complex firms contain a large number of components (or VRs) in continuous
interaction with one another, within and outside the firm (Section 4.2.3).
Understanding the firm’s VRs architecture, from both a structural and dynamic
standpoint, and the conditions set by the outer environment for the firm’s goal attainment, is
thus a first key step for an effective mapping of the firm-system.
Another important aspect to consider is understand how the outer and inner
environment interact and, more particularly, how the firm uses an adaptive behavior —given its
limited resources and capabilities— to adjust to its ever-changing outer environment.
Those researchers and practitioners who attempt to accomplish the Stage 1 tasks should
note that dividing inner from outer environment when mapping the firm-system may have
substantial advantages (Simon 1996):
1. we can assume, up to a certain point, that complexity of the firm is largely a
reflection of the complexity of the environment in which it operates.
2. the firm is designed to create and exchange value as a reflection of the environment
in which it operates.
3. we can discriminate between internal and external complexity, and assess whether
they mismatch or are aligned, and
4. we can more easily come to understand and predict the firm’s adaptive behavior
from our knowledge of the firm’s goals, and its inner and outer environment. In
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
82
other words, if we know what the goal of the firm is, we can seek to envision how its
behavior will change if we modify a few characteristics of the outer environment,
given that we have a reasonable knowledge of the firm’s inner environment.
All the above implies that when mapping the firm-system, it is wise to thoroughly
characterize the structural components of the firm (VRs) leading to value creation and value
exchange first, and thereafter to continue with a comprehensive view of the static and dynamic
interactions taking place between components (VRs). This way researchers and practitioners
might advance more readily toward realism, filling the gap between the mechanistic view of the
firm provided by conventional theories of the firm, and complexity of real-life firms.
B. Activities
The proposed activities covered in Stage 1 (Fig.5) are called to help researchers and
practitioners generate a thorough understanding of the firm’s complexity constituents, ascertain
how they operate and connect, and set the ground for a sound knowledge of complexity:
Figure 5. The CBVF Methodology: Stage 1 Activities
Source: own elaboration
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
83
B.1 Source and Measure Value
According to the previously defined theoretical constituents of the CBVF (Section 4.2),
“value” is the basic building block of the firm and the fundamental element from which the
firm’s components originate and become cohesive. Furthermore, “value” explains the most
inner structure of economic events and substantiates the very nature of firms’ life.
However, as crucial as “value” is for our understanding of the firm, it still remains an
elusive notion that refers to different phenomena, ranging from the equivalence with “price” to
more sophisticated definitions (Haksever, Chaganti et al. 2004). Thus, clarifying the term “value”
from the beginning is key when mapping the firm-system. This should be better accomplished
not disguised as a theoretical disquisition, but by examining the firm value system dynamics,
namely specifying the multidimensional processes through which each particular firm creates
and exchanges value (Fig.6).
Usually the first step in any firm’s value creation process (VCP) is creating Potential Use
Value (PUV), also known as intrinsic value. PUV is created through new combinations of
resources motivated by new opportunities detected by the people in the firm to realize value for
the customer (Ghoshal, Moran 1997). PUV thus attaches the greatest importance to the
customer, as the key driver of value creation, and requires a logic of processes and activities
through which the firm directs the transformation of value inputs.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
84
Figure 6. The Multidimensional Value System Dynamics
Source: own elaboration
PUV is value that remains “dormant” or unrealized in the firm until it is exchanged and
becomes realized. It is at this moment where Exchange Value (EV) comes into the picture. EV is
the monetary amount realized at a certain point in time when the exchange of a new good,
product, service, or task takes place —the amount paid by the user to the seller (Lepak, Smith et
al. 2007)— or the primary mechanism through which previously created PUV becomes realized.
EV usually substantiates through a Value Offering (VO), namely a bundle of benefits
made of goods, services, information, access to systems or infrastructure, and risk-sharing
formulae (Ramírez 1999), or a hybrid combination of them, procured either by the firm alone, or
in mutual collaboration with the customer. Note that the later would make value sourcing even
more complex, because it would need to consider a multiplicity of values in relations with
multiple actors that cannot be reduced to a single metric (Dean, Ottensmeyer et al. 1997).
For EV to materialize there must be a coincidence between an opportunity perceived by
the firm and a potential demand from a customer. In other words, EV depends on a customer’s
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
85
subjective evaluation of the PUV created by the firm. This is what we call Use Value (UV), which
refers to the specific qualities of the VO, as perceived by the customer in relation to his/her
needs (Bowman, Ambrosini 2000, Haksever, Chaganti et al. 2004). As such, UV is perceived by
the customer and cannot be judged in isolation from the wider needs and economic
circumstances of the customer.
In combination, these four dimensions of value —PUV, EV, VO, UV— highlight the
subjective and context-specific nature of the value system dynamics. It is worth pointing out to
the fact that different customers may arrive at different conclusions about his/her UV,
depending on their individual needs or the context in which they are embedded (Lepak, Smith et
al. 2007). Consequently, the kind of potential value that is created (PUV), how it is perceived as
valuable (UV), how it is realized (EV), and the means through which value is exchanged (VO), are
likely to vary considerably depending on the firm and its context.
Altogether mapping the firm-system calls not only for understanding and measuring
what the customer is willing to pay for (UV), but also for measuring the multiple dimensions of
value which co-exist simultaneously. Note that this process greatly differs from quantifying the
financial value of the firm (Koller, Goedhart et al. 2010), or assessing the shareholders’ wealth,
or the value linked to the stakeholders (Hillman, Keim 2001, Harrison, Wicks 2013).
Ultimately, researchers and practitioners should note that what it really matters most
when sourcing and measuring value under the CBVF perspective is to recognize the ability of the
firm to create and exchange value on the benefits that are important to the customer
(Kothandaraman, Wilson 2001). This shall leave in the background relevant managerial decisions
as to what competencies and capabilities are used by the firm, or how the firm captures value.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
86
B.2 Cluster Value Hierarchically
The CBVF approach considers “hierarchy” a central defining constituent of complexity of
the firm, capable of offering insight into many complex phenomena (Clauset, Moore et al. 2008).
Furthermore, the CBVF assumes that value creation is configured in a hierarchical scale, where
top level groups (VRs) divide into secondary groups that further subdivide into lower level
groups, and so forth over multiple levels.
Clustering value, by using hierarchical analysis techniques, can be a productive and
efficient way that we can use to determine the firm’s elemental VRs. This type of analysis is
based on the assumption that non-linear interactions take place in the inner and outer
environment of the firm, which in turn leads to the formation of different levels that are
hierarchically ordered one below another.
Hierarchical levels thus reflect some structural properties that can be very helpful when
mapping the firm-system. The following are some good examples (Helbing 2012):
While changes on the lowest hierarchical levels are fastest, changes on the highest
levels are slow.
On the lowest level, we generally find the strongest interactions among elements;
this would explain why the fast changes occur on the lowest hierarchical level.
Elements do not behave always individually, but form units representing the
elements of the next level. The interactions within this units are stronger than the
interactions between different units.
The relatively weak residual interactions between the formed units induce a
relatively slow dynamics.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
87
The highest hierarchy levels usually take a strong influence on the system on a
relatively short time scale. This also makes it difficult for the lower, less central
levels of the system, to adjust themselves to a changing environment.
As outlined above, a general interdependence between the strength of the interactions,
the changing rate, and the formation of hierarchical levels can be found in most complex
systems. Moreover, the existence of different hierarchical levels implies a separation of scales
(Helbing 2012), which could reasonable be applied to the hierarchical value clustering of the
firm.
From a practical perspective, researchers and practitioners should be aware that the
hierarchical structure of the firm’s value system may help generate meaningful and predictive
hypothesis for higher levels without knowing much about the lower levels. This is particularly so
in the case of less complex firms, where higher levels VRs can provide reliable explanations of
the behavior of the firm, while the analyst only has a broad picture of the VRs below, or no
picture at all.
B.3 Determine VRs Configuration
The previous activities covered in Stage 1 shall have provided researchers and
practitioners with a general perspective of the fundamental value system dynamics of the firm,
and how value is grouped in different levels and dimensions. Now we move one step further and
try to decompose the firm’s value clusters into authentic VRs.
Specifically, VRs share the properties of symbol systems (Simon 1996), thus having the
means to:
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
88
a) collect value-related information from the inner and outer environment.
b) encode it into the firm’s internal activities, resources and/or processes.
c) arrange the internal structure through a value logic, and
d) create unique and differential value offerings that are exchanged within the firm’s
inner and outer VRs.
When determining the firm’s VRs configuration, the analyst should preferably gather all
previous information on the firm’s multidimensional value system, as to generate a
comprehensive description of where value lies in the firm, and of the network linking the
different VRs with the inner and outer environment.
It is important that the analyst is sure about the questions he/she is asking and that
he/she focuses on the key variables (Rooke, Molloy et al. 2008). Regardless of the knowledge
the analyst may have about how the firm works or is organized, he/she should be aware that
what complexity tells us is that patterns can emerge where we least expect them, and that
preconceived ideas and personal believes are bad advisors when determining the appropriate
firm’s VRs configuration. Moreover, different people may have different mental models and,
consequently, will view the firm —and its corresponding VRs configuration— in different ways.
Furthermore, the analyst should not forget that he/she stands before a complex system
where non-linearity dominates and that “strange” or counter-intuitive behaviors may appear.
The later often result from very complicated feedback loops in the system, which can produce
errors in our analysis and undesired side effects (Helbing 2012). For example, in large complex
firms sometimes large organizational units might not have substantial effect on value creation,
while small units might significantly impact value creation. In other occasions, the analyst might
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
89
be tempted to assimilate VRs to decision-makers or organizational units, which most often
would result in fuzzy outcomes.
To avoid these major pitfalls it is important that analysts consider the way in which
mapping the firm-system is to be used. Though history and past experience, for instance, can be
of great help in identifying VRs, not less important is that the analyst devotes enough time to
understand the self-organizing principles of the firm when determining the VRs configuration.
Last but not least, a thorough determination of the VRs configuration should be realized
through the assessment of two other important complexity’s defining characteristics: structural
homeostasis and specialization. The former relates to the mechanisms held by the firm to
maintain the state of some important inner properties within narrow limits before changes in
the outer environment. The later provides an understanding on the number and different type
of unique value related tasks that are accomplished by the firm in its struggle to remain
competitive and meet the customer needs. Both properties are key factors for thoroughly
mapping the firm-system, the knowledge of which should preferably be captured by the analyst.
B.4 Visualize and Analyze the Network
“Network analysis, as a methodological approach, has been one of the great success
stories in the last two decades” (Levinthal 2007). Notwithstanding the foregoing, different
authors provide contrasting definitions of the term “network” upon attending to the different
purpose of their investigations. For example, a strategic view of networks considers a network
as a “long term purposeful arrangements among distinct but related for-profit organizations that
allow those firms in them to gain or sustain competitive advantage” (Jarvenpaa, Ives 1994); from
a business perspective a network could be defined “a set of two or more connected business
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
90
relationships, in which each exchange relation is between business firms that are conceptualized
as collective actors” (Cook, Emerson 1987); still some authors emphasize the differential
informational advantages of participants and the control benefits actors can generate by being
advantageously positioned within a network (Gulati 1998).
From our CBVF standpoint, what is key is that a “network” presupposes a structure of
participants, processes of interaction and exchange, and a unifying purpose among participants.
Moreover, the functions of a network are always wide and diverse and they can be
characterized not only with respect to its activities (efficiency), actors (self-interest) and
resources (leveraging heterogeneity), but also in relation to learning, innovation and resource
development (Anderson, Hakansson et al. 1994).
Upon completion of the previous activities in Stage 1, researchers and practitioners
should be in a good position to graphically represent the firm’s network context and network
profile (Anderson, Hakansson et al. 1994).
The network context is the part of the network that is relevant for the operation of the
firm, and encompasses a direct network —where value is directly exchanged between a dyad of
VRs— and a secondary network —which comprises those VRs that are indirectly connected to
the exchange dyad and thus affect the direct network. The network profile refers to the position
of the firm’s VRs in the network, based on the intensity and strength of the links, the type of
links, the role/s played in the network, and the power relative to other VRs.
A good network representation of the firm should permit researchers and practitioners
to answer questions such as: How are VRs organized in the direct and secondary networks? How
can I tell what the firm’s network looks like when I can't actually look at it? Which VRs in the
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
91
network prove most crucial to the network's connectivity? What approximate effects might
occur if those VRs were removed or failed to create and exchange value? What percentage of
VRs or links among VRs need to be removed to substantially affect the network connectivity in
some given way?
In addition, greater potential may be derived when we provide a firm’s network
complexity macro-perspective (Brandes, Raab et al. 2001), which in turn might enable us to:
1. visualize the overall hierarchical structure of the network more intuitively.
2. capture the relative status of the different firm’s VRs.
3. analyze what and who causes the firm to be hierarchical.
4. compare between different possible network alternatives.
5. understand the value creation and exchange processes within the inner and outer
environment.
Notwithstanding the work involved, analysts should consider the firm’s network
construction as a serious analytical endeavor that may lead to highly productive, descriptive and
normative outcomes. No less important would be the valuable insights for the knowledge of the
firm that might be provided with beneficial implications for managerial practice (Gulati 1998).
C. Tools and techniques
Researchers and practitioners seeking to comprehensively map the firm-system need
methods and tools that help them accomplish the activities involved accordingly. The linear,
mechanistic, causal approaches of the past cannot meet this challenge. Therefore, new methods
and tools based on the principles of non-linearity and networks are essentially required to
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
92
manage the interdependencies of the firm’s complexity constituents on which the CBVF
approach fundamentally relies.
The proposed methods and tools needed to thoroughly map the firm-system are
summarized as follows:
Classification tools.
Hierarchical clustering.
Network visualization techniques.
Value network construction and analysis tools.
Analysts may choose to use a single tool at a time, or a combination of them, when
attempting to accomplish the activities covered in Stage 1. Nonetheless, the use of a specific
tool or a combination of them may depend on the particularities of the firm, the past experience
with the CBVF approach, and the level of proficiency of the analyst using non-conventional tools.
C.1 Classification techniques
These are tools aimed at identifying, analyzing, and classifying the static and dynamic
capabilities of the firm leading to the creation of value in all its four dimensions ―potential
value, use value, exchange value, and value offerings (see Section 4.3.5, B.1). Through the use of
diverse classification criteria, researchers and practitioners should be able to determine how
each source of value in the firm:
integrates into a particular VR.
requires a unique combination of resources.
creates new opportunities for further adding new tangible and intangible value.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
93
is exchanged with other VRs within the firm’s network context, and
converts one type of value to another.
Classifying the sources of value should enable us to assess each value output
individually, along with the activities, resources and processes involved. In doing so, the analyst
might address three different types of questions aimed at assessing, i) the dynamics of the value
creation system of the firm, ii) the exchange of value within the network context of the firm, and
iii) the impact that each VR has on the value system of the firm as a whole. The following are
some examples of the questions that the analysts should attempt to answer (Peppard, Rylander
2006):
Questions about the sources of value creation: How does the firm organize, create,
and bundle value (either by adding new value, combining value internally or
externally, or converting one type of value to another)? Who participates in the
value creation process? With what resources?
Questions about the exchange of value: What is the overall pattern of (internal/
external) value exchange within the firm’s network context? What VRs are more
critical in the exchange of value? What VRs exchange value above or below
expectations?
Questions about the impact of value: What approximate impact does each source of
value have on VRs? How does a VR impact on another VR, and on the firm as a
whole?
While conducting such inquiry on the value system of the firm, the analyst might be
required to use typical business process re-engineering (BPR) procedures, as well as techniques
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
94
such as SWOT analysis, flowcharting, mind mapping, brainstorming, cost-to-benefit analysis,
Delphi method, data survey techniques, data mining, big data analytics, etc. not to say his/her
past experience on the operation of the firm’s value system.
Moreover, it is always advisable when performing a value classification field study to log
all the data captured into distributed data bases, in order to be of help later when completing
the mapping activities. Table 1 below shows an example of a typical outcome obtained from
using value classification techniques.
Value output
From
Dimension of
value
To
Value
Repository
Hierarchical
level
Value
Repository
Hierarchical
level
Distribution
agreement
Alliances
3
Exchange value
Capacity
management
4
Electronic
ticketing app
Customer
experience
4
Potential value
Innovation
5
…
…
…
…
…
…
Table 1. Value Classification Technique
Source: own elaboration
As for the table above, the first row identifies a distribution agreement as a value output
flowing, as exchange value, from the “Alliances” value repository to the “Capacity management”
value repository, the later belonging to a hierarchical level 4.
By classifying this multidimensional information of the firm’s value system, the firm
builds an organized knowledge bank that is key to support later activities within the CBVF
methodology. Moreover, analysts might also want to consider drawing a journey map featuring
the flow/path followed by each value output individually; for example, depicting the process
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
95
followed from the moment a value input is received by a VR and processed, to the moment in
which a value offering containing that original input is exchanged with an end customer.
Analysts considering using classification techniques should be aware that access to the
firm’s data systems is required. Upon access to data, the analyst most probably will find a lack of
correspondence between the firm’s actual data ―usually centered on accounting and
measuring financial performance― and the data required for true value classification. At this
point, the analyst may have no other choice than either to adapt the available data to the
dimensions required by the mapping activities, or to develop new sources of data as he/she
gains more experience with the CBVF (read more in Section 4.3.4).
C.2 Hierarchical clustering
When mapping the firm-system, researchers and practitioners may want to use
different techniques for inferring the hierarchical structure of the firm’s value system and,
subsequently, translate the resulting knowledge into the topological properties of the firm’s
value network before getting insight into more complex network phenomena (Clauset, Moore et
al. 2008).
In hierarchical clustering the analyst seeks patterns by grouping multivariate data into
clusters. The goal is to find an optimal grouping for which the value-related data within each
cluster are similar, the rest of the clusters being dissimilar to each other. From a practical
perspective, the analyst should try to find the groupings of value that make sense in terms of
the firm’s actual behavior and past experience. The groupings so detected are the basis of the
firm’s VRs, on which the next stages of the CBVF methodology heavily rely.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
96
The number of ways of partitioning a set of items into clusters is given by (Rencher,
Christensen 2012):
Hence, hierarchical methods permit us to search for a reasonable solution without
having to look at all possible arrangements.
Unlike other classification techniques, in hierarchical clustering neither the number of
groups, nor the groups themselves are known in advance. The analyst can typically use an
agglomerative method, which starts with clusters, one for each observation, and ends with a
single cluster containing all observations. At each step, an observation or a cluster of
observations is absorbed into another cluster (Rencher, Christensen 2012). The reverse process,
known as divisive method, might also be plausible in pursuit of our goal.
Whatever the hierarchical clustering method that is finally used, a decision needs to be
made as to the optimal number of value clusters found. For this purpose the analyst can
generate a dendrogram and select a number of clusters from the dendrogram, cutting across
the branches at a given level of the distance measure used by one of the axes (Clauset, Moore et
al. 2008, Rencher, Christensen 2012). When determining the value of that provides the best fit
for the data, one approach might be to look for large changes in distances at which clusters are
formed. In this case the analyst should choose the number of clusters with the largest change in
distance.
To check the validity of the resulting cluster configuration, the analyst should test the
hypothesis either supposing there are no clusters or groups in the population from which the
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
97
sample was taken, or through a cross-validation approach where data are randomly divided into
two subsets and , and a cluster analysis is carried out separately on each of and . The
results should be similar if the clusters are valid.
The general tendency of the firms to form tightly connected clusters could be reflected
by using a Clustering Coefficient . This concept is well-known in sociology, where notions
such as “cliques” and “transitive triads” have been widely employed. In the case of the firm, a
might be calculated using the percentage of pairs or neighbors that are themselves
neighbors (Fagiolo 2007).
Another less visible though no less valuable use of hierarchical clustering may be the
prediction of missing interactions. For example, given an observed but incomplete value system,
the analyst might want to generate a set of hierarchical random graphs that fits the firm’s value
system. The analyst might then look for pairs of nodes that have a high average probability of
connection within these hierarchical random graphs, but which are unconnected in the
observed network. These pairs would most likely be the candidates for missing connections
(Clauset, Moore et al. 2008, Jackson 2008).
C.3 Network graph visualization
Visualizing the firm as a network is key in helping researchers and practitioners
understand value-related data, facilitate the graphical analysis of data, and communicate that
understanding to others (Freeman 2000, Brandes, Raab et al. 2001, Jackson 2008).
Given its relevance for mapping the firm-system, we should first start choosing one
among the set of different graph drawing algorithms available —i.e. one based on points and
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
98
lines, matrices, etc. This is a crucial task, since different visual approaches can lead to different
findings that are unlikely to be revealed using non-visual means of analysis. Moreover, different
images or graphs can not only emphasize important features of the structure and dynamics of a
network and react differently upon possible alterations (Strogatz 2001, Barabási, Bonabeau
2003, Moody, McFarland et al. 2005) —i.e. structural complexity, network evolution,
node/connection diversity— but they can significantly affect the scope of the CBVF analysis and
its potential outcome.
In view of the above and given the large number of visualization options offered by
graph theory today, the analyst should carefully address three main aspects to convey the
meaning of the graph quickly and clearly: the content to be visualized, the type of graphical
design, and the algorithm realizing it (Brandes, Raab et al. 2001).
This author suggests a network image based on points (nodes) and lines (arcs, edges),
where points represent VRs and the lines represent the exchange of value between VRs. Also
important is that the analyst takes into consideration a variety of aesthetic criteria; for example,
planarity and the display of symmetries are highly desirable, as well as to keep the number of
bends and crossings low (Battista, Eades et al. 1999).
The network image should allow the firm to focus on the nonlinear dynamics of the
nodes (VRs), without being burdened by additional complexity in the network structure itself.
This can be accomplished if the analyst draws a network graph that is static. Given that most
graphs do a poor job representing change in networks (Moody, McFarland et al. 2005), this
simplification would allow the firm to avoid the issues arising out of structural complexity and to
concentrate instead on the network (more interesting) dynamics. The analyst will need to
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
99
consider the use of a graph drawing algorithm to produce a graph that is easy to understand and
reflects all its possible linkages in an appropriate manner.
Notwithstanding the above, as researchers and practitioners gain experience mapping
the firm-system, a more realistic approach that combines dynamical and structural complexity
of the network should be sought. In this regard, recent media advances should allow analysts to
use space to represent distance and movement to represent change over discrete units of time
(Moody, McFarland et al. 2005).
With these new tools, the analyst might reflect the rate of change in value exchange,
the sequence, or the richness of a value exchange structure. However, as interesting as they
might seem, this practical methods for dynamic network visualization —e.g. network flip books,
dynamic network movies— remain at a very theoretical stage of development and should only
be used by the analyst with great caution.
C.4 Value network construction
Over the years different tools and techniques have been developed by authors trying to
extend the analysis of the firm away from the perspective of an isolated unit to looking at how
the firm creates value within the context of a network. For some of them, constructing and
analyzing the value network has become one of the key guiding forces for determining how a
firm should be improved or developed (Peppard, Rylander 2006).
The following list contains a summary of the main tools and techniques used by
researchers and practitioners to construct a firm’s value network in recent years:
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
100
Value chain (Porter 1985). Although Porter’s value chain approach has been
superseded by more holistic value-based networks, some authors still consider that
the value chain can be a valuable analysis tool for the identification of firm-level
competitive strengths and weaknesses, especially in industrial organizations
(Stabell, Fjeldstad 1998). From a CBVF perspective, the interest of the value chain
lays in acknowledging its influence over the more recent value-based analysis
frameworks, and as an interim value analysis tool that might be used by firms en-
route towards much realistic value networks. Unlike value chain focus on single
firm’s core processes and sequential activities, value networks address more
complex value dynamics and make it possible to see beyond the conventional
resource-based boundaries of the firm.
Value constellation (Normann, Ramirez 1993). The idea of value constellation
builds on the idea that the firm is the center of a constellation of services, goods and
design; one constellation in which customers are also suppliers, and suppliers are
also customers. The value constellation logic presents the firm with three
implications according to its authors: (1) value occurs not in sequential chains but in
complex constellations, therefore the goal of the firm is not so much to make or do
something of value for its customers as is to mobilize customers to create value for
themselves; (2) as potential offerings become more complex and varied, so do the
relationships necessary to produce them, with the most attractive offerings
involving customers and suppliers, allies and business partners in new combinations,
instead of the single firm providing everything alone; and (3) if the key to creating
value is co-producing value offerings that mobilize customers, then the only true
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
101
source of competitive advantage is to make the value-creating system work.
Normann and Ramirez value constellation approach implies that the firm must
continuously reassess and redesign its competencies and relationships in order to
keep its value fresh and responsive.
Value Net (Parolini 1996). Parolini’s main focus is on value-creating systems, which
she defines as a set of activities creating value for the customers. According to this
author, customers not only are the receivers of value, but also play an active role in
value creating processes, together with other actors involved. Parolini’s Value Net
approach aims at identifying all the activities creating value for the customer, and
then to analyze them structurally. Only after the first phases are completed, the
author suggests to go on and consider who does what. The resulting representation
takes the form of nodes and edges, where the nodes represent sets of activities and
related resources which are best considered together, and the edges describe
relations between nodes that can represent flows of goods, information, or financial
resources, depending on which aspect the analyst seeks to develop. In general, the
value net approach provides a practical framework which enables the analyst to
decompose the firm into a sub-set of activities within a value-creating system,
represent the linkages that interconnects them, and guide the implementation of
the methodology step-by-step.
Other value network frameworks. Value networks have been studied for a few
decades now and many researchers have developed their own models and methods
to describe and analyze value creation in networks (Herrala, Pakkala et al. 2011).
Although still at a conceptual stage and with empirical work yet to be done,
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
102
researchers and practitioners focusing on the CBVF might get insight from the
following approaches and use them as a complement to the aforementioned
approaches:
o “Value-adding partnerships” (VAP) (Johnston, Lawrence 1988) shows how a
set of independent companies work together to manage the flow of goods
and services along the entire value-added chain.
o Kothandaraman and Wilson’s value-creating networks are mainly focused
on firms in a network aimed at delivering value to the final consumer. This
model moves beyond VAPs, where firms collaborate to improve their
position in the markets, and builds upon the value-creating network whose
objective is to create superior customer value. Kothandaraman and Wilson
develop a rationale for value-creating networks based on superior customer
value, core capabilities and relationships (Kothandaraman, Wilson 2001).
o Allee’s approach to the enterprise as a living system (Allee 2002). Given that
Porter’s value chain model is inadequate to understand the complexities of
value, Allee proposes that organizations operate according to the principles
of living systems and goes to define three criteria to model businesses and
enterprises: pattern of organization, structure, and process or exchanges.
From Allee’s standpoint, the “molecular level” of economic activity is the
exchange, not only material exchanges but also the intangible. At the end,
Alle defines organizations as “patterns at exchanges”.
o Holweg et al.’s value grid approach describes firms moving beyond
conventional linear thinking and suggests a value grid framework with three
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
103
dimensions —vertical, horizontal and diagonal— aimed at discovering a
variety of new pathways to improve performance (Holweg, Pil 2006). By
mapping the value grid and using a complex and dynamic perspective, the
authors propose to rethink the organization’s value proposition and its
associated structures.
o Herrala et al. propose four key building blocks that impact the value chain of
a product or service: customer value, core competences, relationships and
interactions (Herrala, Pakkala et al. 2011). These building blocks might be
re-used by researchers and practitioners to create a method of analysis on
value creation in networks.
o Helander’s four phases might also be used by researchers and practitioners
to assess and understand value creation in a network by determining: (1)
who is the customer? what customer considers valuable?; (2) what activities
are needed to create the value for the customer?; (3) what resources are
needed to carry out the activities?; and (4) who (actors) are able to utilize
these resources? (Helander 2004).
Researchers and practitioners should note that the best way to advance in the proper
construction of the firm’s value network is to build their own hybrid method, which draws on
the specificities of the firm. In this regard, the tools described above are a good starting point to
reflect on the value creation processes of the firm and might provide some interesting
methodological clues as to later on design a comprehensive and readable value network of the
firm.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
104
4.3.6 Stage 2: Visualizing Complexity
Having completed the activities referred in Stage 1, researchers and practitioners should
now have a well-grounded perspective on the firm as a complex system, as well as a unique
understanding of what complexity looks like in a particular firm.
During the previous stage of the CBVF methodology, a detailed knowledge on the
constituents of complexity shall have been gathered and, more specifically, on the firm’s
structural components and interactions. Furthermore, we should now be in a position to
visualize the complexity of the network representing the firm, as well as some of the
optimization mechanisms developed within the network, such as the specialization and
homeostatic mechanisms.
But before we attempt to visualize firm’s complexity at once, we may wonder what is
that we are interested in visualizing complexity of the firm? And more importantly, what may
we get in return?
The advantages of visualizing complexity of the firm are many and varied. Some of the
most important are introduced right below:
Enable researchers and practitioners to conclude whether the firm is really a
complex system and, therefore, if the activities, tools and techniques proposed by
the CBVF methodology are applicable and can be used.
Some visualization techniques enable us to assess how complex the firm really is.
Eventually, we may want to know how complex the firm is in relation to another
firm, thus comparing firm A with firm B (De Toni, Nardini et al. 2001), or
alternatively, compare the complexity of the firm A at two different points in time.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
105
Make it possible to compare different VRs configuration alternatives, and choose
the one most suited to improve the performance of the firm (Deshmukh, Talavage
et al. 1998).
Assist planners in managing desired levels of complexity, depending on the changing
operating or environmental conditions (Deshmukh, Talavage et al. 1998).
Help determine the degree of fitness (or degree of adaptation) of the firm with
respect to its outer environment, and therefore to plan for managerial guidance if a
misfit occurs.
However, developing a measure for visualizing complexity of the firm that considers all
different aspects of complexity is challenging, even more to grasp complexity in a single
measure or number (Boschetti 2008). In addition, nearly all intended measures of complexity
known to this author come from the fields of physics, biology, or computing science, and they
hardly seem to capture what we intuitively expect from such measures (Adami 2002), nor do
they form part of any theory telling us when or how things get complex (De Toni, Nardini et al.
2001).
Furthermore, the literature usually distinguishes between deterministic and statistical
measures of complexity, and if we focus on the methodology, between computation theory
(Kolmogorov, Rissanen, Universal Turing Machine-based theories) and information theory
(Shannon’s entropy, thermodynamic depth). These approaches are incomplete measures for
visualizing complexity of the firm, sometimes because they do not consider the structure of the
system, other times because they are simply incomputable or limited by the possibility to get
information about the system (De Toni, Nardini et al. 2001). In order to overcome these pitfalls,
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
106
a set of measures that capture researchers’ and practitioners’ intuitive ideas about what is
meant by complexity (and simplicity) of the firm, would be needed.
If we accept that processes in the firm can be viewed as computations, then complexity
of the firm might be regarded as a kind of computational complexity, which would theoretically
allow, in turn, to infer complexity from an appropriate “finite state firm” that reproduces the
(value creation) logic of the firm —something to a certain extent similar to a “Universal Turing
Machine” in the context of the firm.
Aside from such idealization, unachievable in practice, complexity of the firm cannot as
yet be captured simply by attempting to characterize its underlying dynamic value creation
logic, but we should also consider its underlying structural or functional complexity (Adami
2002). Here again difficulties arise when attempting to count the number of different functions
that the firm can perform. Furthermore, it is troublesome to imagine that a “universal” measure
for structural or functional complexity can be devised given that firms differ so greatly in form
and function.
In this challenging theoretical-practical context, it may be helpful if we first identify the
essential characteristics to be considered for visualizing complexity of the firm:
It should intuitively match researchers’ and practitioners’ expectations of what
complexity is (Adami 2002).
It should reflect the dynamics of the underlying value creation processes, as well as
the firm’s structural architecture.
It should take into account information about the interactions with the environment
in which the firm operates.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
107
It should be inferred from empirical observations (Adami 2002).
It should make it possible to introduce the use of mathematics.
By making sure that any potential measure of complexity of the firm satisfies the
characteristics above, we might be able to narrow the search field to only a few plausible
options. Notwithstanding researchers and practitioners should note that the measures that
might help them visualize complexity are, to some extent, context-dependent or even subjective
(Gell-Mann 1995), and will mostly depend on the firm’s own network context and the level of
detail achieved when mapping the firm-system.
A. Objective and scope
The main objective of Stage 2 is to address the problematic of visualizing complexity of
the firm, thus to determine first whether visualization of complexity is possible and, if
appropriate, to establish the most convenient way to depict it.
Such visualization should allow researchers and practitioners to draw conclusions on the
structural components of the firm and its dynamic relationships and, at the end, inform
managerial practice. For example, a particular visualization of complexity of the firm might allow
managers to figure out where “slow” VRs are located, whether there exist inadequate
connections/ interactions that make it difficult for the firm to coordinate adaptive responses in a
timely manner (Eidelson 1997), or whether there are bottlenecks within the firm’s network
context.
A particular set of measures well fitted to visualize complexity of the firm has to do with
network complexity. Network complexity has grown to become the “universal language” to
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
108
describe dynamic evolutionary systems (Barabási 2002). As such, it opens opportunities for the
introduction of new methods for characterizing systems complexity, not as information-based
complexity but, most essentially, as topological complexity of the network representing the
system.
Information-theoretic and non-information-theoretic network complexity approaches
have proven useful to solve many interdisciplinary problems ―e.g. problems in biology,
computer science, ecology, neuroscience, linguistics, sociology, mathematical psychology
(Dehmer 2011), with many authors having contributed various measures of network complexity.
Some of these measures offer great potential for analyzing complex networks quantitatively
and, by extension, provide researchers and practitioners with an effective way to visualize
complexity of the firm.
For a network complexity measure to be deemed as acceptable ―and since no measure
will never provide an absolute and uncontestable account of complexity― it must be sensitive
to changes in one, or a combination of any, of the CBVF’s theoretical constituents (Section 4.2).
More specifically, it should reflect changes in:
the number and variety of VRs in the network.
the hierarchical nestedness of VRs.
the connectivity of the nodes in the network.
the interactions required to exchange value.
Consequently, the scope of Stage 2 should not only cover the selection and later
calculation of a network complexity measure (or a combination of them), but also cover the
study of the results obtained and the subsequent planning for improvement.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
109
Furthermore, it is highly recommended that, as an extension of the later, researchers
and practitioners devote time to study the costs associated with higher/lower than expected
complexity, the effects of different VRs configurations on the variability of the measures of
network complexity, and the relation between such measures and system performance
(Deshmukh, Talavage et al. 1998).
B. Activities
The activities proposed in Stage 2 aim at assessing alternative methods of network
complexity, quantitatively calculate them, analyze the outcomes, and plan for improvements in
the firm-system. The figure below illustrates the order in which these activities might be carried
out (Fig.7):
Figure 7. The CBVF Method: Stage 2 Activities
Source: own elaboration
B.1 Integrate Knowledge
Before we start even thinking on what and how to visualize complexity of the firm, it is
important that we integrate all the structural and process-based knowledge available from
different parts of the firm.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
110
Particularly important is to put together all the knowledge generated in Stage 1, which
should allow us to draw the firm’s network context, as well as to challenge our understanding of
complexity upon testing some preliminary hypotheses about how the firm behaves before
particular shocks. The knowledge so gathered should be made available to those members of
the firm actively involved in carrying out the Stage 2 activities, as well as those responsible for
planning improvement actions.
B.2 Assess Methods and Choose Optimal
Upon having internalized the knowledge on the firm’s complexity constituents,
researchers and practitioners should undertake a review of the theoretical-practical foundations
of the different topological network measures available, with a particular focus on those
methods most related to the firm’s particular structural and dynamical properties mapped in
Stage 1.
Specifically, the analyst should look at the best balance between the set of potentially
usable methods of network complexity and its “calculability” in terms of robustness and cost. At
this point the analyst should take extra care, because the firm will exhibit very different
behaviors in different areas of its “parameter space” (Boschetti 2008), thus significantly
affecting our understanding of complexity and the outcomes resulting.
B.3 Calculate and Analyze Outcomes
Analysts should proceed to calculate the set measures chosen in the previous phase,
and carefully log the results in a data base for later analysis. A brief description of the equations,
the list of parameters, and the values used, should always be provided by the people carrying
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
111
out the measurements. Once calculations are performed, analysts should work to detect any
inconsistencies, bifurcations, or contradictory values within the measurement space,
reconfiguring the set of measures if measurements appear incompatible or are unintelligible.
After completion of calculation tasks, a study of the levels of variability and of the
tradeoffs between different structure-complexity-performance network setups (Deshmukh,
Talavage et al. 1998) should provide a valuable inquiry into complexity of the firm.
A proposal for further enhancement of the network complexity methods and its
calculation process, including those concerned with the analysis of the outcomes, the
communication patterns and associated costs, are all tasks expected to be carried out during
this phase.
B.4 Plan for Improvement Actions
A plan sketching the actions for improving the firm’s network value system and firm’s
performance is a by-product of the activities covered in Stage 2. The Plan should include, at
least, the general vision and goals of the firm, together with a detail specification of the scope,
calendar time, expected outcome, follow-up metrics, and resources required to successfully
implement each individual improvement action.
It is worth noting that although taking improvements into actions is not considered
within the scope of the CBVF methodology, the proper implementation and follow-up of each
action individually should increase our understanding of complexity of the firm and, therefore,
strengthen the feedback between complexity, improvements, and firm’s performance.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
112
C. Tools and techniques
Networks have been extensively characterized both structurally and quantitatively by
graph theory, which has over 150 years of extensive development and application (Bonchev,
Buck 2005). This provides researchers and practitioners with a large number of tools and
techniques to substantiate visualization, particularly by using measures of topological
complexity of networks representing the firm.
A good starting point may be figuring out whether the firm’s network shows the
attributes of a small-world or a scale-free network. These well-known graph topologies imply
particular dynamic properties (M’Chirgui 2012) which in turn may have important managerial
implications in terms of designing better and more robust (value system) networks.
Small-world networks are characterized by shorts paths and high clustering, that is,
they show small average path length —or average shortest path between nodes— compared to
the number of nodes, and a high degree of clustering compared to a random graph of the same
size. Small-world networks display enhanced signal propagation speed —even between distant
parts of the system―, synchronizability and computational power (Watts, Strogatz 1998,
Strogatz 2001, van Ham, van Wijk 2004), and they can help modeling firms’ value systems that
are hard to visualize using conventional graph theory techniques.
Scale-free networks are dominated by a relatively small number of nodes that are
connected to many other nodes. These highly connected nodes dominate the topology of the
network forming hubs. Highly connected hubs become more connected over time and, as a
result, the centrality of these nodes functions as an attractive element for new nodes to join the
network (Barabási, Bonabeau 2003). What it is interesting about scale-free networks is that they
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
113
behave in certain predictable ways thoroughly known by researchers. For example, they are
remarkably resistant to accidental failures, but somewhat vulnerable to “attacks”.
Finding whether the firm’s network meets the properties of a small-world or a scale-free
network would therefore have two important implications (Watts 2004):
1. Very complex networks could be captured by rather simple models, and
2. Well-known/tested metrics and models used to address complex social networks
problems might be applied to the firm as well.
Besides the proper characterization of the firm’s network as a small-world or a scale-
free network, other network parameters can provide further insight into topological complexity
and may be addressed by researchers and practitioners. In this regard, the network analysis
literature provides extensive references of metrics that might be conveniently used for a
thorough descriptive analysis of topological complexity (Wasserman, Faust 1994, Dean,
Ottensmeyer et al. 1997, Jackson 2008, Kolaczyk 2009). Some of the key metrics to be used at
this stage may include:
Nodes characteristics: degree, degree distribution, centrality, including closeness,
betweenness, eigenvector.
Edges characteristics: edge betweenness centrality, hubs, authorities.
Network cohesion: subgraphs, census of cliques, number of dyads and tryads,
coreness, motifs, density, transitivity, reciprocity, connectivity, etc.
Graph partitioning: modularity, hierarchical clustering, dendrogram.
Assortativity and mixing.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
114
Additional measures would comprise others based on the idea of graph entropy and
graph invariants, including Rashevsky’s topological information content (Rashevsky 1955) ,
Mowshowitz’s symmetry index for graphs (Mowshowitz, Dehmer 2010), Bonchev’s indices
based on the combined use of the adjacency and distance matrix (Bonchev, Buck 2005), and
network structural interpretations —e.g. branching in trees, linear tree complexity, cyclicity in
graphs, etc. (Jackson 2008).
To provide an idea of inter-component associations in complex networks, measures
such as the cycle coefficient and others focused on the distribution of connectivity of nodes (Rao
Raghuraj, Lakshminarayanan 2006), may also be useful. This type of measures might even be
used to characterize the relationship between the number of cycles and the robustness of the
firm’s network.
Ultimately, analysts should be aware that all measures mentioned above are useful
within the concrete scope in which they are defined and applied. Thus, a given measure
featuring different firms might behave differently and even contradictorily. The same might
occur if the map of complexity changes, or if we have different observers for the same firm
(Tarride 2013).
4.3.7 Stage 3: Modeling and Simulation
Despite their complexity, firms have many structural and functional features in common
that can be effectively modeled and simulated using advanced computing techniques and
software tools. By performing diverse modeling and simulating tasks, researchers and
practitioners can explore the nature of the firm and their dynamical behavior under a range of
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
115
assumptions. This ability to model and simulate the behavior of the firm is call to have a major
influence on studying and understanding firm’s complexity.
Modeling (the firm) is central to our very understanding of real-life behavioral
phenomena and is probably one of the most consistent approaches to theory construction
(Jaccard, Jacoby 2010). However, now that we have gathered knowledge on the fundamental
structure and logic underlying the basic behavioral phenomena of the firm in the previous stages
of the CBVF methodology, there is still a long way to go until we are able to devise a model. That
is precisely the objective pursued in this stage.
In accomplishing the modeling and simulation tasks set forth below, many variables
need to be assessed. This in turn requires tools and techniques that enhance the integration of
data, and allow interdisciplinary model development, reproducibility of models, and
visualization of the results (McGarvey, Hannon et al. 2004).
Upon building a model of the firm, researchers and practitioners should also be able to
understand what is fixed and what changes in terms of structure, anticipate how value is
created and flows within and outside the components (VRs) of the firm, how and why
interactions between VRs occur, as well as how to address the fundamental tradeoffs among
efficiency, effectiveness and agility (Rouse 2007). The ultimate objective being to identify
potential ways in which to improve the firm’s value system design, before acting on improving
the performance of the firm.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
116
A. Objective and scope
The main objective of Stage 3 is to create an “imitation” of the structural components
(VRs) and interactions shaping the behavior of the firm, as well as to test the imitated system in
a variety of simulated environments to get insight into that behavior. Eventually, our goal is to
achieve as broad as possible understanding of the behavior of the firm and build a certain
predicting capacity.
The activities covered at this stage, as well as the tools and techniques herein
suggested, include powerful means of approaching the real-life behavior of the firm. Whereas
the earlier stages of the CBVF methodology provide us with key knowledge describing the
behavior of the firm, at Stage 3 we aim at building a model in the computer to work out the
implications of a large number of VRs ―the firm’s network context― continuously interacting
one another.
At this point, researchers and practitioners should have accepted the idea that since
trying to explain and anticipate the behavior of the firm in all its particularity would be simply
inaccessible, we need to work via simplified, and sometimes fuzzy variables. After all, we have
no choice but to focus on a few properties abstracted from reality ―i.e. the own fuzzy notion of
value― and disregard the incommensurable number of internal and external variables that
characterize the inner and outer network context of the firm.
Consequently, central to the modeling and simulation activities in Stage 3 of the CBVF
methodology is the process of design itself, which in turn involves carefully delineating the logic
of the firm that influences its performance goals. After this logic is made explicit, the analysis
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
117
and assessment of constraints and alternatives of action becomes key in carrying out our
modeling tasks.
At the end, a final design and simulation against a broad range of conditions should
complete the scope of Stage 3. The outcomes obtained from the simulation should allow us to
corroborate a number of intuitions gathered from previous stages in the CBVF methodology,
draw attention to less obvious relationships, and suggest possible lines of empirical research to
confirm and expand on the insights gathered (Carrillo-Hermosilla 2015).
B. Activities
The activities covered in Stage 3 are grouped around the ODD protocol ―developed by
Grimm et al. (2006)― and Helbing’s (2012) principles for crafting agent-based models. The
reason for settling the activities using a framework specifically designed for use in (ecology and
social) agent-based models, is because it is a method for systematically carrying out complex
systems modeling and simulation activities and is fully consistent with the goals and method
pursued by the CBVF.
The ODD protocol is based on the experience gained by twenty eight experienced
modelers from within the complexity modeling community. The protocol is made up of three
main blocks, namely: Overview, Design concepts, and Details, which are further divided into
seven elements: purpose, state variables and scales, process overview and scheduling, design
concepts, initialization, input, and submodels.
The logic behind the ODD sequence is as follows: (1) context and general information is
provided first (Overview), (2) followed by strategic considerations (Design concepts), and (3)
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
118
more technical details are provided (Details) (Grimm, Berger et al. 2006). Building on the ODD
logic, listed below are the key activities for modeling and simulation of the firm under our CBVF
approach:
1. Purpose: This block informs why the firm needs to build a model, what the firm is
going to do with the model, and what is the purpose of the simulation. The context
and purpose of the model should help researchers and practitioners understand
why some aspects of complexity are included while others are ignored.
2. Variables and scales: The full set of variables is chosen and described at this point.
The term variables refers to the properties of the model’s key components —i.e. the
firm’s VRs. Once the analyst knows the full set of variables, he/she would have a
clear idea of the model’s structure and resolution. Thereafter, the higher-level
components should be also described, for example, if a higher cluster of VRs exist.
Finally, in addition to the variables, the scales addressed by the model should be
also described —i.e. the size of the firm’s network context. Choosing the scale is a
crucial decision which affects the design of the model itself.
3. Process overview and scheduling: Consists in providing a description of the
underlying processes or fundamental mechanisms leading to the particular firm
behavior that needs to be explained, as well as the scheduling of the model
processes. Moreover, this refers to the order in which the processes are performed
and the subsequent order in which the variables are to be updated.
4. Design concepts: The design concepts process provides a framework for designing
and communicating the model. A short checklist of design concepts would include:
emergence, adaptation, fitness, prediction, interaction, observation.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
119
5. Initialization: This process deals with questions such as: What are the initial values
of the variables (or input vector)? Is initialization always the same, or should it vary
among simulations? Are the initial values to be chosen arbitrarily or based on
specific data?
6. Inputs: As the dynamics of most complex models are driven by some environmental
conditions (inputs or constraints) that change over time and/or space, the analyst
needs to know what input data are to be used, how data are generated, and how
data can be captured.
7. Submodels: All submodels representing the processes listed above are presented
and explained in detail. The analyst should also provide a description of the
mathematical “skeleton” of the model and a full model description where the
assumptions are verbally explained. Furthermore, questions such as: What specific
assumptions are underlying the model? How are input values chosen? Or, how are
the submodels tested?, should be answered now.
8. Calibration: The analyst compares the results from the computer simulation with
the empirical evidence, pointing out what features are correctly reproduced and
which not. Moreover, the analyst describes the limitations of the model, as well as
its explanatory power. Finally, the analyst should choose the model with the better
predictive power, namely the one that better matches the data that have not been
used for calibration.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
120
C. Tools and techniques
In this section we propose a set of tools and techniques that support the general
objective and scope pursued by the modeling and simulation activities established in Stage 3.
These tools and techniques include a number of soft computing techniques consisting of several
computing paradigms ―including neural networks, fuzzy cognitive maps, agent-based
modeling― which in turn can be used to produce hybrid modeling systems for solving firm
complexity problems.
The use of one tool or another may depend on one or more of these factors: (1) the
analyst’s previous knowledge of the firm, (2) the particular goals pursued by the model and the
adequacy of abstraction offered by a specific tool, (3) the type, quality and robustness of data
available, and (4) the particular skills of the modeler and the type of software technology
available.
The use of one tool or another is not exclusive. This means that the modeler may choose
to combine several soft computing techniques in order to tackle the complexity and high
dimensionality of real-life firm’s problems.
Hybrid techniques can have different architectures, each having a different impact on
the efficiency and accuracy of the outcomes. For this reason it is very important to optimize the
architecture design (Castillo, Melin et al. 2006). For example, the architectures might combine,
in different ways, neural networks and/or fuzzy cognitive maps and/or agent-based modeling
(Jang, Sun et al. 1997, Panwai, Dia 2005, Rodin, Querrec et al. 2009, Stula, Stipanicev et al. 2010,
Song, Miao et al. 2010, Zarandi, Hadavandi et al. 2012, Lee, Lee et al. 2013), in order to achieve
the ultimate goal of approaching the behavior of the firm.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
121
It is worth noting that the choice of any particular tool or hybrid technique, not only will
determine the level of abstraction and the method to follow by the modeling and simulation
activities (Izquierdo, Ordax et al. 2008), but will also affect the breadth and depth of the
outcomes to be obtained at Stage 3 of the CBVF methodology.
C.1 Artificial Neural Networks
Artificial Neural Networks (ANN) result from attempts to mimic certain aspects of the
information processing and physical structure of the brain through a large number of relatively
simple and interconnected neurons (Caudill, Butler 1992, Li 1994, Kriesel 2007). The underlying
concept assembles many single simple processors (neurons), which run in parallel, learn from
experience, and interact through a dense web of interconnections (Fig.8). These systems are
capable of performing tasks that are found to be extremely complex and difficult by today’s
computing systems (Quaddus, Khan 1999).
Figure 8. Neuron with R Inputs
Source: Hagan, Demuth et al. 2014
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
122
Typically, a neuron has more than one input. The individual inputs are each
weighted by corresponding elements of the weight matrix (Hagan,
Demuth et al. 2014).
The neuron has a bias , which is summed with the weighted inputs to form the net
input :
This expression can be written in matrix form:
where the matrix for the single neuron case has only one row. Now the neuron
output can be written as:
At present there are more than thirty different families of ANN being used in research
and/or industry applications (Krycha, Wagner 1999), though most of them can be grouped in
two basic network paradigms: the supervised and the unsupervised learning networks. ANN in
these two categories differ in their architecture, training, mode of operation, and interpretation
of outputs (Lippmann 1987).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
123
Supervised neural networks mainly include the multilayer perceptron (MLP), recurrent
associative networks, and Hopfield networks. Unsupervised neural networks cover networks
such as adaptive resonance theory models and self-organizing maps. According to numerous
field studies (Vellido, Lisboa et al. 1999, Quaddus, Khan 1999) the vast majority of ANN
practitioners rely on the use of the feedforward MLP trained by back propagation, with only a
few documented investigations opting for unsupervised models.
Over the last decade, and especially in the last few years, ANN have reached into a wide
range of applications, which can be divided into the following main categories: pattern
classification, forecasting, planning, approximation, generalization, and optimization. The special
features of ANN’s information processing makes them attractive for solving complex problems
in finance, specifically in forecasting, trading, stock performance, and portfolio selection;
marketing, including market segmentation, monitoring of customer behavior patterns; retail:
forecast of sales, inventory, staffing, pricing; telecommunications: customer churn, win-back,
assignment of calls, optimal network design, efficient routing, control of traffic; insurance:
detection of fraudulent claims, prediction of claim costs; operations management: scheduling,
control, planning; and many other examples (Vellido, Lisboa et al. 1999, Smith, Gupta 2000,
Paliwal, Kumar 2009).
The experience gained with ANNs and the large body of research carried out on business
and management applications, makes them a sound alternative to conventional statistical
models. In fact, much of the comparative literature points out that ANNs can be applied to many
problems that are solved conventionally by statistical and management science techniques, and
that they outperform, or perform as well as, the conventional statistical models (Vellido, Lisboa
et al. 1999, Krycha, Wagner 1999, Paliwal, Kumar 2009).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
124
Authors frequently cite the following advantages of ANNs (Li 1994, Warner, Misra 1996,
Paliwal, Kumar 2009):
Suitability to handle incomplete, missing or noisy data.
Being a non-parametric method, not requiring any a priori assumptions about the
distribution and/or mapping of the data.
Their parallel processing ability, such that each neuron in the ANN acts a processing
element similar to a Boolean logical unit, except that the neuron’s function is
programmable.
Their distributed memory, which means that a ANN does not store information in a
central memory, thus relying on the collective outputs of all the connected neurons.
Their learning/training ability, which makes ANNs capable of applying learning rules
to develop models while adapting the network to the changing environment.
Their demonstrated capability to approximate any continuous function.
But ANNs also have some important disadvantages. They are not a general-purpose
problem solvers. They are good at solving systems of linear or non-linear equations, organizing
data into equivalent classes and adapting the solution model to environmental changes.
Therefore, they are not so good at logical inference. ANNs do not use a structured methodology
for choosing, developing, training, and verifying an ANN, thus users of ANNs must conduct
sensitivity analyses to identify the best possible configuration of the network. Last but not least,
most ANNs are not able to explain how they solve problems (Li 1994, Warner, Misra 1996,
Vellido, Lisboa et al. 1999, Paliwal, Kumar 2009).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
125
From a CBVF perspective, ANNs are a powerful tool that can be very helpful at the early
stages of the CBVF modeling process. They offer an approach to modeling which does not
require a complete algorithmic specification. Moreover, ANNs provide inductive means for
gathering, storing, and using, experiential knowledge (Schocken, Ariav 1994).
There is a descriptive proximity between ANNs and the firm as an adaptive complex
system. For example, real neuron networks are inspired by the architecture and behavior of the
brain and tend to “fire” only when its combined input exceeds a certain threshold. Not very
differently, the firm under the CBVF approach is comprised of a network of VRs, each of which
continuously interacts exchanging value when certain conditions are met.
From a computational perspective, ANNs can be used to extract a set of rules that
specify how the firm has structured in the past and exchanged value. Literature shows, for
example, that feedforward networks are good at classifying fuzzy objects into concrete
categories (Desai, Crook et al. 1996, Setiono, Thong et al. 1998, Atiya 2001, Smith, Gupta 2002,
Prevolnik, Škorjanc et al. 2011). Assuming that the analyst has gained access to a large amount
of past data and that he/she is able to discern a good set of attributes, ANNs might help him/her
minimize classification error when mapping featuring the firm’s VRs and value exchange
patterns. Additionally, ANNs would help us avoid falling into the consistency trap of expert
knowledge.
The inductive experience gained through ANNs might enable researchers and
practitioners to monitor the firm’s value creation and value exchange patterns over time, so that
they can detect non-performing VRs or identify outputs which seem to be off-target (Schocken,
Ariav 1994). In doing so, ANNs might provide a framework to systematically identify
improvement opportunities in the firm and approach the impacts on firm’s performance.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
126
Last but not least, ANNs might also help us bring order to the data management
processes needed to successfully implement the CBVF approach in the firm. ANNs would make
possible to create a distributed memory organization of data, where data is spread across the
network in a distributed format and would allow retrieval of information through inexact or
incomplete keys. This flexibility would enable managers and analysts to trace chains of
associations and recognize patterns in unpredictable ways (Schocken, Ariav 1994, Smith, Gupta
2002).
Ultimately, ANNs and conventional statistical tools should not be viewed as competing
modeling approaches, but rather they should work synergistically, conventional methods
providing good starting points for ANNs (Sharda, Wang 1996, Krycha, Wagner 1999). In this
regard, literature shows well-performing associations, for example, between feedforward MLPs
and discriminant analysis and regression, and between unsupervised networks and cluster
analysis (Paliwal, Kumar 2009).
C.2 Self-Organizing Maps
Self-Organizing Maps (SOMs) are a data analysis and visualization technique invented by
Finish Professor T. Kohonen designed to provide a way of representing complex, arbitrary
multidimensional data in much lower dimensional spaces, usually in one or two dimensions
(Kohonen 1998, Kohonen 2013, Bhowmick, Shah 2015). This technique is essentially a data
compression process, also known as vector quantization, which creates a network that stores
information in such a way that any topological relationships within a training set are maintained.
One of important aspect of SOMs is that they learn to classify data without supervision.
Unlike supervised training techniques such as backpropagation, in which the output is compared
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
127
with a target vector, training a SOM does not require a target vector ―no right answers are
provided. A SOM learns to classify the training data without any external supervision, making it
possible to obtain insight into the topographic relationships of data. An unknown input is
classified according to a node of a regular, usually two-dimensional grid, the model of which is
most similar with it in some metric used in the construction of the SOM. Every input data item
shall then select the model that matches best with the input item, and this model, as well as a
subset of its spatial neighbors in the grid, shall be modified for better matching (Kohonen 2013).
According to Mangiameli et al. (1996), if we consider a Kohonen one-dimensional array
of neurons, each of which receives the same input vector , the index measures the
dimensionality of the input vector such that = 1,2, …, , and N Kohonen layer neurons are
indexed by the numbers j = 1, 2, …, n , then any particular Kohonen neuron j has an input
weight vector . The neuron c is the neuron with weight vector that is closest to the
input signal vector . This distance is calculated as follows:
If we define as the subset of neurons that includes and its adjacent neighbors, the
process of self-organization is accomplished as follows (Mangiameli, Chen et al. 1996):
where . The magnitude of the learning coefficient determines how
rapidly the system adjusts over time. Typically alpha is decreased as learning proceeds. The
neighborhood function that defines starts with a large area and decreases over time.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
128
An example that illustrates the way a SOM works is mapping the three dimensional
components of color ―red, green and blue― into two dimensions (ai-junkie ). The Fig.9 below
graphically illustrates how a SOM can be trained to recognize the eight different colors shown
on the right. The colors are presented to the network as 3D vectors, one dimension for each of
the color components, and the network learns to represent them in the 2D space. In addition to
clustering the colors into distinct areas, areas of similar properties are usually found adjacent to
each other.
Figure 9. Example of How a SOM works
Source: ai-junkie
SOMs are widely applied to clustering problems and exploratory data analysis in
industry analyses, finance, biomedical and telecommunications, with practical applications
ranging from industrial process control to the management of very large document collections.
Other more specific applications of SOMs found in the literature include profiling of the
behavior of criminals, categorization of galaxies, categorization of real estates, and linguistics
(Kohonen 2013).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
129
From a CBVF perspective, SOMs might be used as a clustering and data segmentation
technique for highly multidimensional firm networks, and as a substitute for hierarchical
clustering methods (Mangiameli, Chen et al. 1996). As SOMs compress information while still
preserving the most important topological and metric relationships of the primary network data,
they make it possible to know which neuron provides output when we are not so much
interested in the exact output of the neuron (Serrano-Cinca 1996, Smith, Gupta 2002, Kriesel
2007). For example, firms with a high-dimensional network context or VRs configuration could
be mapped onto a SOM, thus generating a two dimensional discrete grid topology that it might
be more easily visualized. This way researchers and practitioners might find easier to identify
neighborhood relationships and check whether optimal behavioral areas (clusters) develop. In
this sense, SOMs might be used as nonparametric similarity graphs, or clustering diagrams.
C.3 Bayesian Networks
Bayesian Networks, also known as belief networks, are a class of graphical models that
represent the joint probability distribution between a given set of variables or entities of
interest. From the CBVF perspective, these entities of interest can be thought as nodes
representing the VRs of the firm’s network, the edges featuring their associations (relations),
and the Bayesian network showing the joint probability distribution between the entities of
interest.
Formally, Bayesian networks are directed acyclic graphs (DAG) whose nodes represent
random variables in a Bayesian sense ―they may be observable quantities, latent variables,
unknown parameters or hypotheses― and edges represent conditional dependencies. Each
node is associated with a probability function, whose input is a set of values for the node's
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
130
parent variables, and the output the probability distribution of the variable represented by the
node (Nagarajan, Scutari et al. 2014).
Bayesian networks has its roots in Bayesian statistics and focuses on the computation of
posterior probabilities or densities. For a given a set of variables , a
Bayesian network thus allows a concise representation of a DAG, , where each node
corresponds to a variable . The correspondence between the graphical separation ()
induced by the absence of a particular edge and probabilistic independence () provides a
convenient way to represent the dependencies between the variables. Such a correspondence is
formally known as an independency map (Pearl 2014). The correspondence between the
structure of the DAG and the conditional independence relationships it represents, is
explained by a directed separation criterion, or d-separation (Nielsen, Jensen 2009, Pearl 2014).
Fig.10 below illustrates the annotation for the fundamental connections in Bayesian networks.
Converging connection
For A
Serial connection
For A
Diverging connection
Figure 10. Fundamental Connections in Bayesian Networks
Source: (Nagarajan, Scutari et al. 2014)
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
131
Efficient algorithms exist that perform learning and inference in Bayesian modeling
(Ando 2010). The task of fitting a Bayesian network is usually called learning, and it is performed
in two different steps, which correspond to model selection and parameter estimation
techniques in classic statistical models.
The first step is called structure learning and consists in identifying the graph structure
of the Bayesian network. Several algorithms have been proposed in the literature, which usually
fall under three broad categories: constraint-based, score-based, and hybrid algorithms. As an
alternative, the network structure can be built manually from the domain knowledge of a
human expert and prior information available on the data.
The second step is called parameter learning. As the name suggests, it implements the
estimation of the parameters of the global distribution. This task can be performed efficiently by
estimating the parameters of the local distributions implied by the structure obtained in the
previous step (Nagarajan, Scutari et al. 2014).
Bayesian network has drawn attention across a wide spectrum of disciplines that include
biology, medicine, ecology, health care and, more specifically, in causal studies where causal
relations are encoded by the structure (or topology) of the network (Steyvers, Tenenbaum et al.
2003). In particular, Bayesian networks are useful abstractions of the underlying biological
pathways and signaling mechanisms, where they show the ability to discover new associations
in addition to validating known associations between the entities of interest (Grzegorczyk,
Husmeier 2009, Nagarajan, Scutari et al. 2014).
From a CBVF perspective, Bayesian networks might be useful when researchers and
practitioners attempt to integrate multiple interactions, outcomes, and information from VRs,
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
132
with large amounts of data coming from different sources, or when there is missing data and
uncertainty (Chen, Pollino 2012). In this case, Bayesian networks might provide a good
representation of the probabilistic relationships between a concrete VRs configuration and its
associated flow (exchange) patterns, thus allowing reasoning under the uncertainties associated
with these probabilities.
The inference capabilities provided by Bayesian networks might also help us reveal
possible causal relationships between the firm’s VRs, based on certain implicit assumptions,
uncertain data, or the analyst’s prior knowledge on the firm’s value system. The strength of
these relationships would be established as a conditional probability (CP) attached to each VR
(node). CPs would reflect the degree of belief (probability) that a VR is in a particular state given
the states of the parent VRs ―the nodes that directly affect that VR. Evidence would be entered
into the Bayesian network by substituting the a priori beliefs of one or more nodes with
observation or scenario values (Chen, Pollino 2012). This belief propagation would enable
Bayesian networks to be used for diagnostic or explanatory purposes (Castelletti, Soncini-Sessa
2007), as well as for predicting the states of VRs when data is partial or uncertain.
C.4 Fuzzy Cognitive Maps
Fuzzy Cognitive Maps describe the behavior of a complex system by means of graphs
consisting of nodes, so-called “concepts”, that are connected through arrows that show the
direction of influence between concepts. A positive/negative arrow pointing from concept to
concept indicates that concept causally increases/decreases concept . To reflect the
strength of causal links, weights are assigned to the arrows. Each concept represents a state or a
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
133
characteristic of the system, and these concepts interact with each other as to display the
dynamics of the system (Stylios, Groumpos 1999).
FCMs are regarded as a simple form of recursive neural networks (Kosko 1987), where
concepts are equivalent to neurons, but unlike neurons, they are not either “on” or “‘off”
, but can take states in-between that are considered ‘‘fuzzy’’. Fuzzy concepts are
non-linear functions that transform the path-weighted activations directed towards them (their
‘‘causes’’) into a value in or .
When a neuron ‘‘fires’’ (when a concept changes its state), it affects all concepts that
are causally dependent upon it. Depending on the direction and size of this effect, and on the
threshold levels of the dependent concepts, the affected concepts may subsequently change
their state as well, thus activating further concepts within the network.
Fuzzy Cognitive Maps can be formulated in different ways. Stylios and Groumpos (1999)
distinguish three different types. FCMs of Type I calculate the value of each component by
computing the influence of all other components to the specific components. This is done by
calculating the following equation:
(1)
Where is the value of component at time , is the value of
component at time , is the weight of the interconnection between component
and component , and is the sigmoid function:
(2)
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
134
Other squeezing functions are the and
that convert the result of the
multiplication into the fuzzy interval or .
FCMs of Type II take into account the previous value of each concept, so that the last
value of each concept is involved in the determination in the new value of concept and so the
values of concepts will have a slight variance after each simulation step. The mathematical
formulation of Type II FCMs is as follows (Stylios, Groumpos 1999):
(3)
Where is the value of concept at time , is the value of concept at
time , is the value of concept at time , is the weight of the
interconnection from to , and is a threshold function. The parameter represents the
proportion of the contribution of the previous value of the concept in the computation of the
new value and expresses the influence from the interconnected concepts in the configuration
of the new value of the concept . The two parameters and satisfy the equation:
(4)
In FCMs of Type III, a concept can take into account its own past value with a weight
. Therefore, Type III FCMs are close to Type II and the calculation rule will be similar to the
equation (3):
(5)
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
135
Where is the value of concept at time , is the value of concept at
time , is the value of concept at time , is the weight of the
interconnection from to , is the weight with which the previous value of concepts
participate in the calculation of the new, and is a threshold function.
FCMs have several properties that make them particularly useful for modeling and
simulating the firm under the CBVF. FCMs focus on those aspects of a system that are uncertain
and not knowable through simple information gathering. For these uncertainties, they provide a
limited number of possible states, so-called scenarios, whose main purpose is to challenge
prevailing mind-sets and avoid the common problem of over- and under-prediction of change
(Jetter, Kok 2014).
To this end, FCM-based scenario approaches put heavy emphasis on integrating
knowledge from experts and/or stakeholders and on eliciting and communicating assumptions.
The maps so created can be based on interviews, text analysis or group discussions and be easily
modified or extended by adding new concepts and/or relations or changing the weights
assigned to causal links.
To attenuate some of the limitations and weaknesses of expert-based knowledge, a
number of learning algorithms have been developed mainly consisting of modifying the FCM
weight matrix. These learning techniques have developed on the following three directions
(Papageorgiou 2012): (1) the production of weight matrices on the basis of historical data, (2)
the adaptation of the cause–effect relationships of the FCM on the basis of experts’
intervention, and (3) the production of weight matrices by combining experts’ knowledge and
data. The learning algorithms resulting from these learning paradigms are Hebbian-based,
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
136
population based, and hybrid ―combines the main aspects of Hebbian-based- and population-
based-type learning algorithms. These learning algorithms are the most efficient and widely
used to train FCMs.
In addition to the learning algorithms above, several FCM extensions have been
proposed during the past decade aimed at improving the original formulation of FCM made by
Kosko. Some of the extensions reviewed in the literature (Papageorgiou, Salmeron 2013)
include: Rule-base Cognitive Maps, Fuzzy Grey Cognitive Maps, Dynamical Cognitive Networks,
Fuzzy Cognitive Networks, Evolutionary Fuzzy Cognitive Maps, Fuzzy Rules incorporated in Fuzzy
Cognitive Maps, to mention only a few.
Further example of the growing interest that FCMs arouse is the number of application
areas in which FCMs are applied. Papageorgiou and Salmeron (2013) study on the most common
application domains of FCMs, identify applications in fields such as environmental studies,
medicine, engineering, business and management, mathematics, computer science, and some
others.
In summary, FCMs allow a quantitative and qualitative analysis of the behavior encoded
in FCM models to aid decision making. By means of FCMs, managers can agree on plausible
combinations of input values and calculate the states of the dependent variables, thus assessing
the impact of input variations ―i.e. variations due to particular policies― and/or the impact of
alternative system description ―i.e. due to different mental models belonging to the same
complex problem. The special combination of explanatory and predictive capacity provided by
FCMs makes them a particularly well fitted tool to serve the goals pursued by our CBVF
approach.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
137
C.5 Agent-Based Modeling
Agent-Based Modeling (ABM) is a modeling and simulating tool that has gained growing
attention over the past fifteen years or so, by the increasing number of applications that call for
models incorporating complex elements of human and social behavior (Macal, North 2009). Up
until now these models were considered too complex to be adequately modeled, however, as
our computational power has increased and data is more fine-grained, we are now capable of
modeling using agent-based models.
ABM has its roots in complex adaptive systems, and is comprised of autonomous, self-
contained, interacting agents, with diverse, heterogeneous, and dynamic attributes and
behavioral rules. Moreover, agents are adaptive, which means they can learn from their
environment and dynamically change their behavior in response to their experiences (Macal,
North 2009).
ABM works best when modeling heterogeneous agents’ relationships and agent
interactions that are complex, nonlinear, discontinuous, or discrete (Bonabeau 2002). These are
models commonly used when mathematical models can be written down but not completely
solved, or when writing down equations is not a practical approach (Axtell 2000, Helbing 2012).
Hence, ABM tend to be descriptive rather than normative —seeking to optimize and identify
optimal behaviors.
Nowadays ABM is being applied to many areas, covering social, physical and biological
systems. Applications range from modeling ancient civilizations (Lima, Hadzibeganovic et al.
2009), to modeling customer and market behavior (Twommey, Cadman 2002), population
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
138
dynamics (Pablo-Martí, Santos et al. 2015), or simulate technological change (Carrillo-
Hermosilla, Unruh 2006).
Examples of ABMs found in the literature span from small, minimalist models —with
simple rules of behavior and a set of idealized assumptions that can be varied over many
simulations— to large-scale decision support systems, designed to answer a broad range of real-
world policy questions. Worth noting is how some authors, such as Chang and Harrington, have
even used ABM to explore a new approach to organizations. For these authors, an organization
can be viewed as a collection of agents, interacting with one another in their pursuit of assigned
tasks. The performance of an organization in this framework is determined by the formal and
informal structures of interactions among agents, which define the lines of communication,
allocation of information processing tasks, distribution of decision-making authorities, and the
provision of incentives (Chang, Harrington 2006).
For most researchers and practitioners the main reason for using ABM is because i)
agent-based models can easily and naturally incorporate the complexity arising from individual
behaviors and interactions that exist in the real-world, and ii) ABM can capture emergent
phenomena and become an effective route to develop a thorough understanding of models
with unstable equilibria (Axtell 2000, Bonabeau 2002). Furthermore, since ABM models are
“solved” merely by executing it, when a particular agent-based model, call it , produces result
, one has already established a sufficiency theorem, the formal statement being (Axtell
2000).
ABM is also well suited for visualizing an organization from the viewpoint not of the
business processes but of the activities, which is what most people inside an organization
usually do. In addition, ABM is flexible, because it makes it easy to add more agents to a
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
139
particular model and tune the complexity of the agents. This flexibility allows the analyst to
readily change levels of description and aggregation, with different levels of description
potentially coexisting in a given model (Bonabeau 2002, Helbing 2012). Another advantage of
ABM not very often mentioned is that once a model has been created it provides not only one
aspect of the solution —i.e. the equilibria, or the stability— but rather entire solution
trajectories (Axtell 2000).
Notwithstanding the foregoing, ABM has one significant disadvantage: despite the fact
that each run of the agent-based model yields a sufficiency theorem, a single run does not
provide any information on the robustness of such theorem. To address this problem, the
analyst must run the model multiple times, systematically varying initial conditions or
parameters in order to assess the robustness of the results (Axtell 2000).
In addition, sometimes an idiosyncrasy in the rule code may produce an output that we
could erroneously take as a significant result of the model. This might happen, for example,
when the agent interaction methods impose some spurious correlation structure on the overall
population. Although no real solution exist to this problem, aside from careful programming,
one could look for the existence of such artifacts by making many distinct realizations of an
agent model, perturbing parameters and rules.
From a CBVF perspective, an agent-based model may be a model that people in the firm
would acknowledge as useful, as it aims to explain what the firm does to create and exchange
value. This has important implications when it comes to populating, validating, and calibrating
the model, since if people connects to the model, they can help to dramatically improve it, and
quantify more easily what needs to be quantified.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
140
Conceptually, a CBVF’s agent-based model would consist of individual agents
(represented as VRs), each with different states and rules of behavior, conveniently represented
in software as objects. Building such a model would amount to describe an agent population
(the firm’s VRs network context), make VRs interact, and monitor what happens. The essential
idea would be that VRs only interact at any given time with a limited (small) number of other
VRs in the population, thus it is possible to define a local neighborhood with interactions limited
to a small number of VRs that happen to be in that neighborhood.
At the simplest level, a CBVF’s agent-based model would consist of a firm with just a few
VRs and the relationships (exchanges) between them. Even this very simple model could display
complex behavioral patterns and provide valuable information about emergent behavior and
the dynamics of the real-world firm that it emulates. Further on, VRs might be capable of
evolving, allowing unanticipated behaviors to emerge.
For more complex firms, ABM might even incorporate other complementary techniques
such as ANN, evolutionary algorithms, and other learning techniques to allow realistic learning
and adaptation (Bonabeau 2002). Such a powerful set of modeling and simulation tools might
become the first line of attack to the modeling problem on the CBVF approach, with researchers
and practitioners resorting to more conventional mathematical modeling techniques only to
“tidy up” what the agent-based model had clearly demonstrated to be a robust feature of the
problem.
C.6 System Dynamics
System Dynamics (SD) is a general term associated to the study of a variety of complex
systems, which explicitly take into account the dynamic behavior that results due to delays and
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
141
feedbacks in the system (Forrester 1994). Besides a set of conceptual tools that enable us to
understand the structure and dynamics of complex systems, SD is also a modeling method to
build formal computer simulations of complex systems and use them to design more effective
policies and organizations (Sterman 2000).
The method of SD permits the analyst to decompose a complex social or behavioral
system into its constituent components and then integrate them into a whole that can be easily
visualized and simulated. This process is illustrated in a graphical way, assuming that systems
can be represented as a collection of stocks connected by flows (Mendoza, Prabhu 2006, Voinov,
Bousquet 2010). Typically, influence diagrams using nodes and directed arrows are used to
denote this dynamic behavior. Fig.11 provides an example of a causality loop diagram generated
by a focus group (Mendoza, Prabhu 2006).
The principles of SD can be summarized in two major statements:
1. Stocks, flows, and delays determine system behavior.
2. Based on the idea of bounded rationality (Simon 1982), SD does not address all the
variables of a problem, rather concentrates on the ones that are key to the problem
and its context, neither does it pretend to optimize, but to satisfice our
understanding of the problem.
As a modeling tool, SD works best when it is used to develop a model that solves a
particular problem, and not a model as a whole or to gain insight. Furthermore, SD considers
modeling a feedback process covering constant iteration, continual questioning, testing, and
refinement (Sterman 2000), not a linear sequence of steps.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
142
Figure 11. Causal Diagram of Management of a Forest
Source: (Mendoza, Prabhu 2006)
SD is grounded in control theory and the modern theory of nonlinear dynamics,
therefore there is a strong mathematical foundation in the theory and models developed under
SD (Sterman 2000). SD is also designed to be a practical tool that policy makers can use to help
them solve the pressing problems they confront in their organizations and, as such, it has been
successfully applied in a wide variety of business and socio-economic fields to understand the
problems and gain an insight into various policy interventions.
Among the main tools used by SD it is worth mentioning causal loop diagrams ―for
representing the feedback structure of systems― stocks and flows ―for mapping networks of
stocks and flows of a system― and integration of flows and differentiation of the stock ―to
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
143
infer the behavior of the flows from the dynamics of the stock (Luenberger 1979, Sterman
2000).
The use of SD might fit the objective and scope of our CBVF modeling and simulation
activities under some circumstances, especially when our understanding of the firm’s VRs
configuration and the interactions between them is sketchy or vague, and not amenable to a
formalized representation. When this is the case, soft SD is a valuable tool to use, thus
overcoming some limitations of other hard computing models unable to fully represent cause-
and-effect relationships.
Furthermore, the method of SD strongly emphasizes the endogenous behavior of the
system and evaluates the dynamic behavior implied by feedback loops (Forrester 1987,
Izquierdo, Ordax et al. 2008). All the above, combined with the fact that SD focuses on
observable variables and provides the modeler with a graphical description language that
describes the interdependencies between the attributes of the target system (Gilbert, Troitzsch
2005), suggests that it might be an acceptable modeling and simulation tool to use and that it
would meet the goals of our CBVF approach.
4.3.8 Stage 4: Optimization
Given the knowledge gathered so far on complexity of the firm, at this stage of the CBVF
methodology we are likely to be in a position to run some optimization process aimed at
achieving a more efficient and productive functioning of the firm’s network value system.
As we have seen in the previous sections of this chapter, the firm is formed by the
interconnection of many value repositories (firm’s component) that may have different, or even
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
144
opposing, optimization goals one another or from the firm as a whole. Consequently, various
structural and dynamical properties of the firm can be explicitly involved in the optimization of
specific functions (Motter, Toroczkai 2007).
In particular, network optimization has a long research tradition within the field of
mathematical programming, and is one of the principal ongoing areas of research in the
optimization arena. This is mainly due to the abundance of network applications and the
confluence of mathematical theory and computer science research (Meyer 1985).
Examples of real-life network optimization problems include finding shortest paths,
finding the vulnerability of networks to disconnection because of link and/or node failure,
maximizing the flow in networks, minimum-cost flows, etc. (Khuller, Raghavachari 2013).
Correspondingly, there are a large number of methods for achieving optimization, including
systems structures and computational methods (Leondes 1998). Last developments in
optimization include non-cooperative mathematical games in decision environments, using
numerical methods such as steepest decent, fixed point, gap function and computational
intelligence algorithms (Friesz, Bernstein 2015).
Given the importance that continuous improvement and optimization presumably have
in the adaptive behavior of the firm, the CBVF approach calls network optimization activities to
play an important role in shaping the evolution of the firm’s behavior at different scales.
A. Objective and scope
Based on the context of optimization described above, there are probably a good
number of behavioral problems of the firm that might be better defined as optimization
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
145
problems, where optimal behavior is most often connected to a function that the system
performs, which is multivariate or multifaceted (Motter, Toroczkai 2007).
Identifying what tendency within the network context improves the behavior of the firm
as a result of a selection pressure imposed, for example, by a customer portfolio is the objective
of Stage 4. Notwithstanding, this goes beyond the conventional definition of optimization
centered on obtaining the solutions that strictly extremize a well-defined functional. Therefore,
instead of trying to find “the” best solution, our CBVF approach prefers to find a “good enough
solution” or a “better solution”, which in turn is more likely to become a workable alternative
for the firm.
According to Motter and Toroczkai (2009), there are fundamentally four major types of
optimization problems related to networked systems, in which the CBVF may focus:
1. Type I: Structural Optimization. Involves finding a graph , where is the set
of nodes and is the set of edges that extremize a given structural functional .
2. Type II: Dynamics Optimization on Static Graphs. For a given graph and a
dynamical system Φ on , such that
(1)
we must find the values of the parameters {α} that extremize a global functional
.of the dynamics Φ. The variables are quantities associated with properties
of the nodes and edges in the network. This type is useful to study flow
optimization problems.
3. Type III: Structural Optimization for Dynamics. For a given dynamical system (1), and
a set of parameters {α} , we must find a graph for which a global functional
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
146
of the dynamics Φ is extremized. This type is useful in design problems, such
as finding the network structure that is optimal for an information outcome.
4. Type IV: Dynamics-Driven Network Optimization. If the graph of the network evolves
in time , we must find the values of the parameters {α} for
which a global functional of the dynamics Φ and of the graph , is
extremized. This type is the most difficult to solve because both the structural
properties and the dynamics can change.
As we can see, optimization in complex networks has broad significance and may
incorporate static and dynamic properties at the same time. The ultimate goal being to serve as
an instrument to analyze (and shape) the evolution of the firm and design actions that influence
the performance and robustness of the firm.
B. Activities
The network optimization activities covered in our CBVF methodology build on the
insights gained from the preceding modeling and simulation tasks performed in Stage 3. From
this point, it is suggested that the analyst carries out the following proposed activities (Floudas,
Pardalos 2013, Khuller, Raghavachari 2013, Friesz, Bernstein 2015):
1. Define the optimization problem for which a solution is to be sought —e.g. graph
theoretic, flow optimization, structural and dynamic optimization.
2. Gather sample data, and choose the parameters to model.
3. Choose a fitness function (or what we want to optimize) and determine when and
how the analyst should evaluate the function.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
147
4. Simulate the model several times and assess the results; determine the procedure
for further exploration.
5. Assess computation time of the simulation; if the simulation requires much time,
use distributed computation.
6. Run the optimization algorithm.
7. Analyze results.
8. Change initial conditions, if required, and run the algorithm as appropriate.
9. Draw final conclusions.
10. Plan for the structural and/or dynamical changes that should improve the network
representing the firm.
11. Start a new cycle of the CBVF methodology by updating the map of complexity of
the firm (Stage 1, Section 4.3.5).
C. Tools and techniques
Many different tools and techniques allow us to achieve optimization in network-like
models, however, not all are thought to deliver the outcomes expected by the CBVF. The
following section describes only a selection of those optimization techniques that might better
serve the requirements of our CBVF.
C.1 Network flows optimization
Network flows optimization has always been a core problem in almost all industries,
including business management. In fact, because network flows optimization problems arise in
so many problem contexts (Ahuja, Magnanti et al. 1995), it is sometimes difficult for researchers
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
148
and practitioners to fully appreciate the benefits and variety of network applications that can
bring to the theory of the firm.
Among the wide variety of network flows optimization problems, the Minimum Cost
Flow Problem (MCFP) is the most fundamental. This problem is generally stated as the “least
cost shipment of a commodity through a network in order to satisfy demands at certain nodes
from available supplies in other nodes” (Ahuja, Magnanti et al. 1993). Special cases of the MCFP
are the Shortest Path Problem and the Maximum Flow Problem.
The study of the Shortest Path Problem (SPP) is a natural point where we may start
optimizing the firm network value system and its underlying VRs configuration. The SPP would
thus refer to the best way to traverse the firm’s network to get from one VR (node) to another
as cheaply as possible. The SPP might typically adopt one or more of the following forms: (1)
finding shortest paths from one VR to all other VRs, when edges lengths are non-negative; (2)
finding shortest paths from one VR to all other VRs in the network; and (3) finding shortest
paths from every VR to every other VR. Typical algorithms used to solve the SPP include the
label-setting and the label-correcting algorithms.
The Maximum Flow Problem (MFP) would typically refer to how we can exchange as
much value as possible between two VRs without exceeding the (interaction/exchange) capacity
of any edge. The generic augmenting path and the pre flow-push algorithms are typically used to
solve the MFP (Ahuja, Magnanti et al. 1993).
The MFP and the SPP are complementary approaches, because they both arise as sub
problems of the MCFP problem. However, the two problems differ because they capture
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
149
different aspects of the MCFP: the SPP models edge costs but not edge capacities, meanwhile,
the MFP models capacities but not costs.
The MCFP usually combines ingredients from the SPP and MFP, so it is very common
that researchers and practitioners end up using linear programing methods to solve MCFP
problems. Among the most popular algorithms used for solving the MCFP are the polynomial
and network simplex algorithms.
As described above, there is a rich set of algorithms for solving network flows problems,
the review of which would be too long for the scope of this thesis. It is therefore part of the
analyst’s work to choose among the most efficient algorithms available for solving the network
flow optimization problem of his/her choice.
C.2 Neural networks global search methods
Neural networks pervasiveness as a tool to approximate unknown functions to any
degree of accuracy, makes them a priori eligible for modeling complexity of the firm. However,
some of the most widely used optimization techniques for training neural networks, such as the
back-propagation technique, are shown limited in its ability to find global optimal solutions
(Sexton, Dorsey et al. 1999). Instead the literature demonstrates that global search techniques
are a far superior technique in providing optimal solutions.
In this section we are going to examine two global search techniques that might serve
adequately the CBVF’s optimization requirements: Genetic Algorithms and Simulated Annealing.
Genetic Algorithms (GA) are a family of computational methods first described by John
Holland in the 1960s and further developed in the 1970s, inspired by the process of biological
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
150
evolution. GAs exploit the concept of evolution by combining potential solutions to a problem,
thus allowing to solve various classes of problems and, more specially, optimization problems
(Holland 1975, Mitchell 1998, Calvez, Hutzler 2006, Joyce, Hayasaka et al. 2012).
By means of GAs the potential solution of a problem is encoded on a linear data
structure, which is called a chromosome. The algorithm works on a set of several chromosomes,
each of which represents a candidate solution to a given problem, called a population.
Recombination and mutation operators are then applied to this population.
Intuitively the role of these operators is to pick up the best part of chromosomes to
obtain a better chromosome. The suitability of the new resulting chromosomes, as solutions to
the given problem, is evaluated by a fitness function. A new population is finally created from
the initial population before starting again the whole process (Mitchell 1998, Calvez, Hutzler
2006, Joyce, Hayasaka et al. 2012). The success or failure of a GA resides on the fitness function,
as it defines the smoothness or roughness of the solution space —namely, the set of all possible
GA chromosomes— and therefore influences the ability of the GA to converge on the best
solution.
GAs have been used widely in applications as diverse as medical image processing,
modeling crystal formation, functional modeling of the human brain (Joyce, Hayasaka et al.
2012) and, more specifically, in a large number of scientific and engineering problems, including
optimization, automatic programming, and machine learning (Mitchell 1998).
In the field of economics, GAs have been used to study evolutionary aspects of social
systems, and to model processes of innovation, the development of bidding strategies, and the
emergence of economic markets (Amman, Tesfatsion et al. 2006). In these applications, GAs
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
151
make it possible to explore a far greater range of potential solutions to a problem than do
conventional programs (Holland 1975).
From a CBVF perspective, GAs might be used in the context of multi-agent models, such
as those specified in Stage 3, as a type of unstructured search method to assist the analyst in the
task of variable selection, determining the optimal topology of a network, or fine-tuning the
model (Vellido, Lisboa et al. 1999, Amman, Tesfatsion et al. 2006, Calvez, Hutzler 2006). For
example, parameter setting in an agent-based model can become long and tedious if the analyst
has no accurate, automatic and systematic strategy to explore the parameter space. In this case
GAs might be used to explore the parameter space and find the best parameter set with respect
to an optimization function. In other cases, GAs might allow the development of new strategies
or decisions that had not been considered in the initial model.
Simulated Annealing (SA) draws from the process of annealing, which occurs when
physical substances, such as metals, are raised to a high energy level (melted) and then
gradually cooled until some solid state is reached to reach the lowest energy state. During this
process physical substances usually move from higher energy states to lower ones, so if the
cooling process is sufficiently slow minimization naturally occurs (Sexton, Dorsey et al. 1999).
Computational SA basically starts at an initial random point, from which the algorithm
takes a step within a range predetermined by the user. The new point’s objective function value
is then compared to the initial point’s value in order to determine if the new value is smaller. If
the objective function value decreases, it is automatically accepted and it becomes the point
from which the search will continue. The algorithm will then proceed with another step. As the
algorithm progresses, the length of the steps declines, closing in on the final solution (Sexton,
Dorsey et al. 1999).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
152
SA algorithms are always able to find the optimum or, at least a point very close to it,
and they are also less likely to fail on difficult functions. Furthermore, SA is largely independent
of the starting values and can escape from local optima and find the global optimum by the
uphill and downhill moves (Goffe, Ferrier et al. 1994).
The literature also shows that SA algorithms perform well in the presence of a very high
number of variables, and when compared with other conventional algorithms on econometric
models, SA algorithms show superior performance (Goffe, Ferrier et al. 1994). Essentially, SA
provides more features at the cost of an increase in the execution time for a single run of the
algorithm. However, when compared to the multiple runs often used by conventional
algorithms to test different starting values, SA are competitive.
Among the frequently mentioned advantages of SA over conventional optimization
algorithms, we may refer the following:
1. SA can escape from local maxima by moving both uphill and downhill.
2. the step length gives the analyst valuable information about the function, so that if
an element is very large, the function is very flat in that parameter.
3. SA is a better overall measure of flatness than a gradient evaluation at a single
point.
4. SA can maximize functions that are difficult or impossible to optimize otherwise. A
drawback of SA is the required computational power, but this problem is gradually
disappearing.
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
153
SA applications include computer and circuit design, the solution of the traveler
salesman problem, pollution control, the reconstruction of crystalline structures, image
processing, and neural networks (Goffe, Ferrier et al. 1994).
It is worth noting that in the case of neural network optimization, some authors observe
that GAs systematically obtain superior solutions than SA (Sexton, Dorsey et al. 1999). Generally,
the GA process of moving from one population of points to another enables it to discard
potential local solutions and also to achieve superior solutions in a computationally more
efficient manner than SA. Furthermore, GAs usually provide researchers and practitioners with
superior estimates of interpolation data.
C.3 Agent-based optimization
Optimizing an agent-based network/model can have different connotations. Most
commonly it refers to how the analyst can influence a network in order to best achieve some
specific goal, while in other contexts optimization may refer to parameters or model design. In
this later case, agent-based optimization can still refer to the optimal choice of a sequence of
external inputs to achieve a particular goal (Oremland, Laubenbacher 2014).
The stochasticity inherent to ABM means that special care should be taken when
attempting to solve optimization problems. As data from individual simulations often vary,
careful analysis of ABM dynamics is a prerequisite for the development of any optimization
technique (Oremland, Laubenbacher 2014). Furthermore, the increasingly number of variables
and complexity of ABM makes it is impossible to enumerate the solution space to find optimal
solutions. As a result, heuristic algorithms are the most frequent means of performing optimal
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
154
control and optimization of agent-based models, thus providing a means for searching the
solution space in an effective manner and in a reasonable timeframe.
Optimization algorithms usually begin with a user’s choice of the control input values
and employ various methods to refine (decrease) the value of the associated cost function, until
no better solution can be found. In addition to the already described GA and SA algorithms,
other potentially useful optimization algorithms include the Tabu search algorithm (Glover
1989), the GRASP Greedy Randomized Adaptive Search Procedures (Feo, Resende 1995), and
Hinkelmann et al.’s method for translating an agent-based model into a polynomial dynamical
system (Hinkelmann, Murrugarra et al. 2010).
CHAPTER 4. COMPLEXITY-BASED VIEW OF THE FIRM
155
4.3.9 Summary
Tools and Techniques
Classification techniques
Hierarchical clustering
Network graph theory
Value network construction
Topological characteristics: Nodes
characteristics, Edges characteristics,
Network cohesion, Graph partitioning,
Assortativity and mixing.
Rashevsky`s topological information content
Mowshowitz`s symmetry
Bonchev`s indices
Others (Raghuraj et al. 2006, Jackson 2008)
Artificial neural networks
Self-organizing maps
Bayesian networks
Fuzzy Cognitive Maps
Agent-based modeling
System Dynamics
Network flows optimization
Global search methods
Agent-based optimization
Activities
1. Source and Measure Value
2. Cluster Value Hierarchically
3. Determine the VRs Configuration
4. Visualize and Analyze the Network
1. Integrate knowledge
2. Assess Methods and Choose Optimal
3. Calculate and Analyze
4. Plan for Improvement
ODD protocol (Grimm et al. 2006)
Principles for crafting agent-based
models (Helbing 2012)
1. Assess results from simulation
2. Run optimization algorithm
3. Analyze results
4. Plan for structural and/or dynamical
changes
5. Start new cycle
Stage
1. Mapping the
Firm-System
2. Visualizing
Complexity
3. Modeling and
Simulation
4. Optimization
Table 2. Summary of the CBVF Methodological Framework
Source: own elaboration
THIS PAGE INTENTIONALLY LEFT BLANK.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
157
CHAPTER 5.
TESTING THE THEORY AND METHOD:
AIRLINES FIELD RESEARCH
5.1 Introduction
Up to this point in the thesis we have been concerned with formulating a complexity-
based theory of the firm and a practical methodology to bring the theory into practice.
Nonetheless, “in social research, generating theory goes hand in hand with verifying it” (Glaser,
Strauss 2009), thus we must make sure that the CBVF is readily understandable and can be
operationalized by researchers and practitioners of the firm.
In accomplishing this key task of our theory building process, the CBVF must provide
clear assumptions that are verifiable in present and future research. This involves that the CBVF
should “fit” the situation being researched, and make it “work” when put into use. By "fit" we
mean that the assumptions made by the CBVF must be applicable to real data; by "work" we
mean that the CBVF must be meaningfully relevant to, or be able to, approach the behavior of
the firm (Glaser, Strauss 2009).
In this chapter we provide a test case for verifying the applicability of the CBVF theory
and method, using field research in network airlines. Field research differs significantly from
other research methods in that it involves various levels of observation, interaction and
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
158
participation with members of real firms. This in turn should provide the researcher with data
against which to confront the CBVF main assumptions and insight to test the theory, which
might not be obtained using other research methods (Young 1999). Furthermore, by examining
the data captured through field research we expect to provide feedback that improves the CBVF
theory and discloses its limitations.
5.2 Field Research Design
5.2.1 Purpose
The purpose of the present field research is to validate the propositions made by the
CBVF theory and to ascertain the coherence and practicality of the procedures established in its
methodological framework. Note that the validation of these propositions must not be confused
with quantitative hypothesis testing; the purpose of field research is to gather and describe
enough data as to verify the applicability of the CBVF theory and method, and not to prove it
statistically.
Based on the data and insights gathered from this empirical fieldwork, verification of
our CBVF theory and method is expected. By iteratively confronting our analytical
generalizations in the form of CBVF propositions with the empirical insights in the form of
validated propositions, internal validity of the CBVF is to be established.
5.2.2 Research questions
The following questions are of particular importance for the present field research:
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
159
RQ1: What are the key value repositories comprising the airlines’ value system?
RQ2: What are the key constraints affecting value creation in network airlines?
RQ3: What is the impact of value repositories and constraints on airlines’ Operating
Margin?
RQ4: What actions help create value in network airlines and affect performance
positively?
Therefore this field research dives into the essence of the airlines’ value system
dynamics with a dual purpose: (1) to offer researchers and practitioners a broad picture of the
external variables and structural components that shape network airlines’ value creation
system, and (2) to address the question of what actions could be more appropriate to increase
value creation in airlines and stay competitive. The main goal pursued by this research is thus of
a qualitative and practical nature, with a particular focus in providing new insights that support
managers in the process of improving value creation and anticipating performance.
It is worth noting that this field research generates novel data from experts on the
constraints and value dynamics within the firm’s network context. Furthermore, we study the
relationships of these elements with a key performance indicator, Operating Margin, as to build
a network model of the firm on the basis of soft computational techniques.
The resulting model, and the outcomes obtained from the subsequent simulation
scenarios, are not a concluding step to solve all the critical competitive issues in the airline
industry. They only provide a proof of concept and set the ground for a novel complexity-based
view of the firm, the generalization of which should open new opportunities to all firm
researchers and practitioners. Hence our intention becomes the exploration, identification and
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
160
description of multiple relationships regarding complexity in airlines, rather than the
quantification of a single consensus or framework.
5.2.3 Why network airlines?
Contrary to the trend of many research studies in the literature, which selectively
choose examples for its confirming power, this author chose network airlines ―also known as
hub-and-spoke airlines― in a non-aprioristic way and without theoretical control. The only
criteria used was past professional background of the author and the fact that the airline
industry is a conspicuous paradigm of a rapidly changing firm, whose many challenges and
opportunities make it one of the most competitive and complex industries of all.
Leading and managing an airline, or one of its operational units, is not an easy job.
Airlines’ managers are invariably compelled to readapt their policies to the customer
preferences as few other industries do, and the ever-changing “external factors” in the industry
impose severe constraints to growth and competitiveness. Complexity has unequivocally seized
an industry that is key for economic growth and progress, where value is increasingly more
difficult to create and where taking the benefits of it, a question of survival (International Air
Transport Association 2011). To make things even more intricate, no magic formula seems
plausible to cope with this complex panorama.
However, among the management options implemented by the airline industry in
history, some pathways seem more likely to succeed than others in the struggle for
competitiveness. One of the pathways that have traditionally yielded better results is when
airlines attempt to solve the competitiveness equation from inside the organization, no matter
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
161
how convoluted the environment becomes. In following this course of action, value creation
seems to become key.
Notwithstanding the foregoing, the notions of “value” and “value creation” are rather
ambiguous and inconclusive, not to say they are full of misconceptions; a kind of slippery floor
for airline managers. With such problematic characterizing the airline industry, the CBVF seems
a natural approach that may help tackle such complex issues and draw conclusions of practical
order.
5.2.4 Method
The method used in this field research follows the stages established in our CBVF
methodological framework, as enunciated in Chapter 4: (1) data capture, (2) mapping the firm-
system, (3) visualizing complexity, (4) modeling and simulation, and (5) optimization.
The author carried out extensive fieldwork in order to capture relevant data on value
creation dynamics from airlines’ experts that could be used to fulfil the goals of the thesis field
research. It should be noted that despite the limited domain of the research (airline industry),
this does not constrain our ability to verify the applicability of our complexity-based theoretical-
practical approach in a real case scenario.
The above involved launching a Delphi process and setting up an Experts’ Panel. After
data were collected, an exploratory data analysis was conducted that contributed to map the
firm-system. Then a network graph of airlines was drawn, the topological complexity of which
was appropriately assessed. Finally, the author undertook specific soft computing modeling and
simulation work, as a step towards reproducing the impact of key variables on the airlines’ value
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
162
system and performance. No optimization work was undertaken as part of the field research as
this exceeded the scope of our field research and the commitment gained from the members of
the Experts’ Panel.
Two more factors characterize the method used in our field research: (1) all inputs,
analysis and modeling tasks performed, as well as the outcomes obtained, are built on the
knowledge and experience provided by an international group of airlines’ practitioners/experts,
and (2) the qualitative and quantitative techniques used are of non-conventional type. The latter
meaning that this author intended to surpass the conventional, cause-and-effect mindset, and
instead applied an approach based on fuzzy elements from reality.
5.2.5 Data collection
As stated before, data on value creation and exchange is key in carrying out our field
research, and a critical asset to make our CBVF theory and methodology both robust and
reliable. However, airlines ―and firms in general― do not generate value-driven data and, if
available, experience shows that managers are typically hesitant to share this information with
others given the “strategic” nature of it. Hence, our first concern in this field research has to do
with how to collect this type of data, and from whom.
In examining the key characteristics of the Delphi method as a survey and knowledge
extraction procedure (Chapter 4), it becomes clear that its philosophy and procedures match
well with our research intentions. First, the Delphi method fits well our research focus on future
actions. Second, it is a method that systematically delivers when relatively little is known about
these actions. Third, our research questions lend themselves to making use of a wide range of
experts geographically dispersed internationally, an area where Delphi excels. Last but not least,
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
163
the Delphi method works especially well when the research consists of an iterative process
where each question builds on the answer to the previous question, and some kind of
consensus is sought (Skulmoski, Hartman et al. 2007, Amos, Pearse 2008). For all these reasons,
the Delphi method was the author’s choice for capturing data.
A. The Delphi process
The broad range of expertise and layering of questions required in this field research
involved a phased approach of four Delphi rounds, each with its own objective yet relevant to
the subsequent Delphi rounds.
Owing to limited access to concrete airlines executives and industry experts, the author
chose to conduct the Delphi process on an online basis, on the belief that our topic of interest
would attract the interest of experts not reachable by other means. This basically involved
launching a Delphi process aimed at attracting worldwide experts to participate; experts who
belonged to many different airlines, held different positions of responsibility, and performed in
various functional areas.
Similar research findings indicate that a Delphi so conducted not only increases the
efficiency of the process, but accommodates experts availability and reduces drop-out-rates.
Moreover, when the online/computer-based Delphi format is compared with a conventional
Delphi, no significant differences between the two are really observed and final survey results
are not seriously affected by changes in the survey procedure (Gnatzy, Warth et al. 2011).
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
164
As the Delphi process has been reviewed elsewhere in this thesis (Chapter 4) and
extensively studied in the literature (Adler, Ziglio 1996, Linstone, Turoff 2002), we exhibit below
an overview of how this author used it in the field research (Fig.12).
Figure 12. The Delphi Process
Source: own elaboration
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
165
Research Questions. These are the research questions derived from the need to
verify the applicability of the CBVF theory and methodology. Previously to the
formulation of the questions, the author considered his own experience in the
industry and the conclusions obtained in Chapter 4 of the thesis. Also a review of
the literature was conducted to determine if a theoretical gap existed.
Research Design. After formulating the research questions, we proceed to select the
research participants, choose a survey and communication system, and draft a
working plan. Participants are a critical component of Delphi research, since it is
their expert opinion upon which the output of the Delphi is based. So a careful
selection of participants was carried out by the author (see Appendix C). The survey
and communication system is also a critical tool in the process, from which an
effective and efficient implementation of the Delphi depends (see Appendix H.3).
The system chosen by the author guaranteed a service level and automated delivery
of questionnaires, thus accelerating the monitoring and reporting of participants
generated data. The working plan is also an essential part of the Delphi process, as it
contains the specification of the roles of researchers and participants, sets the
timeline, and assigns the resources needed to implement the entire Delphi process.
Delphi Round 1. Consisted of a divergent thinking process focused on the
identification by participants of as many value constraints and key firm’s
components as possible.
Delphi Round 2. Aimed at building “experts consensus” on the top ten constraints to
value creation and the top fifteen value repositories identified in the previous
round. This limiting numbers were set to forestall excessive analytical work for
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
166
participants, as well as to avoid an overly too complex network graph and
subsequent model and simulation overload.
Delphi Round 3. Having got an idea of the key structural components making up the
airlines’ value creation system, the focus shifted in Round 3 to the identification of
the connections between the constraints and the value repositories, the
connections among the value repositories themselves, and between the value
repositories and Operating Margin.
Delphi Round 4. Aimed at building “experts consensus” on the questions asked in
Round 3, adding a novel question about the sign of the relationships previously
identified by the experts.
The Delphi process was supported by a website, http://www.valueinairlines.com,
created and hosted by the author to serve as main communication channel with participants,
and which included information about the purpose and method of the research, the research
information sheet file, and introductory resources for participants. Screenshots of the website
can be found in Appendix G.
B. Panel questions
Much of the outcomes in the field research depend on the type of questions posed to
Panel members. The Delphi process used a series of questionnaires interspersed with feedback
(Appendix D), designed to identify the firm’s key components and its relationships. Each
questionnaire in subsequent rounds of the process was drafted based on the results of the
previous questionnaire. The process stopped when sufficient information had been exchanged
and group stability ―defined as the consistency of responses between successive rounds― had
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
167
been reached (Skulmoski, Hartman et al. 2007, von der Gracht 2012). For sake of clarity, and in
order to avoid being too deterministic in the research, specific academic parlance was
deliberately avoided in the questionnaires.
The questions posed to Panel members in Round 1 were twofold: “What do you think
are the main external constraints to value creation in airlines?” and “What are the key value
repositories in airlines that affect airlines’ Operating Margin?” An explanation of what a Value
Repository was provided in the questionnaire along with a link to an author’s introductory paper
on the basics of our complexity-based view of the firm.
Individual answers from Round 1 ―individually compared with the aggregated Panel
answers― were later distributed to each participant in Round 2, and they were asked to agree
on the top ten constraints and the top fifteen value repositories previously identified.
Specifically, the questions posed were: “What do you think are the 10 key constraints to value
creation in network airlines?”, and “What do you think are the key 15 value repositories affecting
performance in network airlines?”
The questions posed in Round 3 pursued to gain insight into the interconnectedness of
the constraints, value repositories and Operating Margin, captured in rounds 1 and 2. In
particular, the questions in Round 3 were threefold: “How do the 10 consensus value constraints
impact on the 15 consensus value repositories?”, “How are the 15 consensus value repositories
interlinked and how they impact on each other?”, and “How do the 15 consensus value
repositories impact on airlines' Operating Margin?” The questionnaire consisted of several
matrix-type questions, where experts could mark whether or not a particular relationship
existed and provide the weight (strength) of a particular relationship by choosing one among
five levels ―zero, very weak, weak, strong and very strong― in a Likert-type scale.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
168
Finally, in Round 4, an individual report containing the answers from Round 3 and a
benchmarking of the personal responses against the aggregate responses from the Experts’
Panel was distributed to the members (Appendix B). Upon the analysis of the report, Panel
members were asked to accomplish the following three types of tasks:
To set the final strength and sign of the links between the 10 consensus value
constraints and the 15 consensus value repositories.
To set the final strength and sign of the links between the 15 consensus value
repositories.
To set the final strength and sign of the links between the 15 consensus value
repositories and airline's Operating Margin.
Similarly as in Round 3, the questionnaire in Round 4 consisted of several matrix-type
questions where experts provided their insight into the strength and sign of the
interconnections among constraints, value repositories and Operating Margin.
All Delphi questionnaires in the field study were conducted using an online Qualtrics-
created survey (Appendix H.3). Copy of the questionnaires used in the Delphi process are found
in Appendix D.
C. Participants selection
The Delphi process required participants to be experts on the phenomenon under
investigation. Therefore, an important practical consideration in the field research design had to
do with who might serve as a qualified expert in the Delphi Experts’ Panel.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
169
The answer to this question was influenced by the particular research questions posed
in the study, and the following eligibility criteria was set for meeting “expertise”: (1) knowledge
and experience with the issues under investigation; (2) position of responsibility held in the
organization, considering that expert competency is usually higher for participants whose
positions are closely associated with the investigated topic; (3) capacity and willingness to
participate, and (4) effective communication skills (Adler, Ziglio 1996).
To decide on the candidates for the Experts’ Panel, the author’s own professional
LinkedIn online network served as the tool used to review candidates’ curricula, and as main
channel of communication with candidates (see Appendix C for a list of Panel members).
On invitation, selected candidates were kindly requested to accept participation in the
research project before starting to provide their insights through the Delphi questionnaires.
D. Criteria for consensus
As described before, the Delphi process used in this field research gathers a
heterogeneous number of international experts from the airline industry, the objective being to
include all diverse opinions and expertise, as to verify the applicability of the CBVF theory and
methodology in airlines. This means that although consensus among participants might be an
outcome of the process, this was not the primary intention of the process (Rauch 1979). Instead,
all the viewpoints were captured with the intention to creatively explore complexity in airlines
according to our CBVF approach.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
170
It should be pointed out that when we say that “consensus” among participants is
reached, what this author means is that a sufficient clarification and definition of the different
opinions and viewpoints among Panel members was effectively reached.
5.3 Results Obtained
5.3.1 Data analysis
In this section we provide an analysis of the data captured throughout the four rounds
comprising the Delphi process, as well as an analysis of data on participation. For sake of clarity,
we divide this section into five subsections: one providing an overall analysis of Panel members’
profile and participation, and one for the analysis of the data captured in each Delphi round.
A. Data on profiles and participation
The total number of participants providing a response at any round of the Delphi
process is thirty three. Notwithstanding that number of participants, we must note that not all
of them produced a response considered technically valid. Consequently, if we focus strictly on
the number of participants providing a valid response ―either complete or incomplete― then
we must adjust the total number of participants to twenty eight.
A.1 Participants by functional area
The professional profile of the participants in the Experts’ Panel is mixed and varied,
covering a broad spectrum of functional areas within their own organizations. Table 3 below
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
171
shows the distribution of the Panel members according to the functional areas where they
perform their responsibilities.
Functional area
No.
Participants
Commercial
9
Network & Revenue
7
Sales & Distribution
5
Corporate
4
Engineering
3
TOTAL
28
Table 3. Delphi Survey, Distribution of Panel Members by Functional Area
Source: own elaboration
As we can see in the table above, Panel members perform their responsibilities within a
wide range of functional areas, with the participants from the commercial area being the most
numerous, followed by those participants performing in the network and revenue areas. This
not only reveals a high functional diversity among participating members, but it also helps us
prevent bias in the field research.
A.2 Members by geography
Geography is another key factor that allows us to characterize Panel members’ profiles.
In this regard, the presence of a broad number of different geographies in the field research
should be a good indicator that reflects consistency and move us away from research bias.
As Fig.13 indicates, participants from Europe are the largest group, followed by
participants from South America, North America and Middle East in this order. Note that this
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
172
geographical distribution pattern remains virtually unchanged across the four rounds of the
Delphi process.
(*) R1: round 1, R2: round 2; R3: round 3; R4: round 4
Figure 13. Distribution of Participants by Geography
Source: own elaboration
Such varied distribution of geographies fulfils a good practice in similar field research
studies, consisting of having a wide geographic coverage of participants to avoid bias. An
explanation for the decrease of participation across the four Delphi rounds (Fig.13) is provided
in the next sections.
A.3 Members by organic position
When analyzing the positions held by Panel members in their respective organizations,
we should take into consideration the heterogeneity existing in the way airlines refer to organic
positions. This sometimes makes it difficult to compare positions or levels of responsibilities in
different airlines, particularly when members are from different companies and/or geographies.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
173
That said, we synthesize the different positions held by Panel members according to the
levels of responsibility stated in their curricula and seeking homogeneity in the description
(Table 4).
Position
No.
Participants
Senior Vice-President
2
Vice-President
15
Director
8
Manager
3
TOTAL
28
Table 4. Distribution of Panel Members by Organic Position
Source: own elaboration
The table above shows that the group of Vice-presidents is the largest among Panel
members, followed by Directors, Managers and Senior Vice-Presidents in this order. This
distribution of organic positions ensures that all data captured in the Delphi process come from
experts with the greatest knowledge and highest experience within the airline industry, as was
originally intended in the design of the Delphi process.
A.4 Type of responses
Panel members provided different types of responses to the questions posed in the
Delphi process, which basically fell into two broad categories: valid and not valid responses.
Valid responses were those responses that could be satisfactorily processed and
analyzed to the effects of the field research, and thus incorporated to the stream of research.
Valid responses were themselves either “complete” ―whenever Panel members provided a
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
174
response to the entire questionnaire― or “incomplete” ―when members responded partially to
the questionnaire.
Not Valid responses refer to those members’ submissions lacking the possibility to be
processed and analyzed by the author, the reason being an incomplete submission process or a
technical failure occurring during submission. Not valid responses were excluded for the
purpose of the field research. Fig.14 shows the distribution of the different types of responses
provided by Panel members along the four rounds of the Delphi process.
(*) R1: round 1, R2: round 2; R3: round 3; R4: round 4
Figure 14. Type of Responses Provided by Panel Members
Source: own elaboration
As we can observe in the figure above, the proportion of “complete” responses mostly
increased across the four rounds of Delphi, at the same time as the proportion of “not valid” and
“incomplete” responses decreased. This trend is consistent with the fact that some Panel
members at the initial stages of the Delphi process might have been hesitant to participate in
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
175
the research, thus preferring to peer into the content of the questionnaires before deciding a
valid submission. This fact might explain most of the “not valid” responses.
Furthermore, the decrease in the number of total valid responses, and of Panel
members’ participation, is consistent with the author’s past experience and numerous Delphi-
based research projects shown in the literature (Linstone, Turoff 2002, Skulmoski, Hartman et al.
2007). From a mere methodological perspective, the total number of valid responses achieved
along the field research is significant enough for an online Delphi process, and ensures a high
level of reliability in the data collected.
B. Delphi round 1
In Round 1 of the Delphi process, Panel members were asked two main questions:
Q1: What do you think are the main external constraints to value creation in
airlines?
Q2: What are the key value repositories in airlines?
Both questions were open-ended, which means that Panel members were free to write
down whatever constraint and value repository they thought were key for real-life airlines. This
process, known as “group visioning” or “divergent thought”, sought to carry out a scattered
creative process where Panel members provided as many different responses as possible.
Given the heterogeneity in the responses obtained from Panel members, this author
synthesized the complete list of constraints and value repositories into a reduced number of
common, non-overlapping categories.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
176
Table 5 below shows the frequencies of the main constraints provided by Panel
members. Note that the top four constraints ―Labor costs, Competition from other airlines,
Government regulation, Fuel cost― amount together to more than 50% of all responses given by
Panel members.
Constraints
Frequency
Percentage
Labor costs
18
15
Competition from other airlines
16
13
Government regulation
15
12
Fuel cost
14
12
Airport fees
8
7
GDS feed
6
5
Leisure travel demand
5
4
Business travel demand
5
3
ATC fees
3
2
IT systems costs and complexity
2
2
Other cited constraints
30
25
TOTAL
121
100
Table 5. Delphi Round 1: Summary of Top Cited Constraints
Source: own elaboration
On the other hand, responses to the question “What are the key value repositories in
airlines?” reveal a very different pattern. As we might expect given the novelty of the term
Value Repository, members’ responses appear much less grouped than in the case of
constraints, with the top four value repositories amounting barely to a 26% of the total number
of responses. Not surprisingly, we also find an evident high level of dispersion in the responses
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
177
(Table 6), which might be due to the broad idea of “value” dominating among Panel members
and their different contexts and backgrounds.
Value Repositories
Frequency
Percentage
Capacity management
6
7.9
Information management
5
6.6
Network
5
6.6
Customer experience
4
5.3
Scheduling
3
3.9
Sales
3
3.9
Procurement
3
3.9
Operations management
3
3.9
Relationships with stakeholders
3
3.9
Corporate culture
3
3.9
Products and services
3
3.9
People and talent
3
3.9
Process and cost optimization
3
3.9
Digital channels
2
2.6
Alliances
2
2.6
Other cited value repositories
25
32.9
TOTAL
76
100
Table 6. Delphi Round 1: Summary of Top Cited Value Repositories
Source: own elaboration
C. Delphi round 2
With the aggregated information from top constraints and top value repositories
gathered in Round 1 in their hands, Panel members were asked in Round 2 to reach consensus
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
178
on the key top ten constraints and top fifteen value repositories making up an airline’s value
system. Specifically, the questions asked in Round 2 were the following:
Q1: What do you think are the 10 key constraints to value creation in network
airlines?
Q2: What do you think are the key 15 value repositories affecting performance in
network airlines?
The analysis of the responses to Q1 and the comparison with the responses given in
Round 1, show how the triad of constraints comprising Government regulation, Fuel cost, and
Competition from other airlines, remain at the top of the list (Table 7). This give us a general idea
of the high degree of consensus reached by Panel members with respect to the key constraints.
Top 10 Constraints
Frequencies
Round 1
Round 2
Government regulation
15
15
Fuel cost
14
11
Competition from other airlines
16
11
Commoditized product offering
1
11
Power of unions/labour force
1
10
Labor costs
18
10
Slot availability
1
8
Excess capacity
1
7
Capital intensity
1
7
Business travel demand
5
7
Table 7. Comparison of Top 10 Constraints, Round 1 vs Round 2
Source: own elaboration
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
179
The consensus reached in Round 2 becomes self-evident when we observe the number
of constraints with hardly any significance in Round 1 that now appear at the top of the table in
Round 2 ―e.g. Commoditized product offering, Power of unions/labour force, Slot availability,
Excess capacity, Business travel demand. An analogous process, but in the opposite direction,
occurs with constraints such as Labor costs, whose frequency drops to the middle of the table
despite being the most cited constraint of all in Round 1.
Meanwhile, the analysis of responses to Q2 and the comparison of the top 15 value
repositories in rounds 1 and 2 (Table 8) reflects a somewhat similar underlying process of
consensus. In this case, some of the most cited value repositories in Round 1 ―e.g. Network,
Capacity management, Customer experience, Information management― remain at the top of
the list in Round 2, despite the large number of different value repositories provided by Panel
members.
Round 2 also shows new entries to the list of top 15 value repositories ―e.g. Brand,
Innovation, Safety and security, Alliances― and the drop of other previously top labelled value
repositories ―e.g. Scheduling, Procurement, Relationships with stakeholders, Digital channels.
This process of entries/exits is a typical characteristic of Delphi groups sharing aggregated
knowledge and building consensus.
Finally, yet importantly, let us recall that the top 10 constraints and the top 15 value
repositories are the only elements that qualify for the next rounds of the Delphi process. All
others, although important, remain excluded from further analysis in the field research.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
180
Value Repositories
Frequencies
Round 1
Round 2
Network
5
17
People and talent
3
16
Revenue management
2
16
Management/Leadership
2
15
Capacity management
6
14
Corporate culture
3
14
Customer experience
4
13
Alliances
2
12
Brand
2
12
Innovation
1
12
Distribution strategy
2
11
Safety and security
1
11
Customer-centric proposition
2
10
Information management
5
10
Process and cost optimization
3
10
Table 8. Comparison of Top 15 Value Repositories, Round 1 vs Round 2
Source: own elaboration
D. Delphi round 3
On the basis of the top 10 constraints and top 15 value repositories determined in
Round 2, the Delphi process continued on to ask Panel members the following questions:
Q1: How do the 10 consensus value constraints impact on the 15 consensus value
repositories?
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
181
Q2: How are the 15 consensus value repositories interlinked and how they impact on
each other?
Q3: How do the 15 consensus value repositories impact on airlines' Operating
Margin?
What follows below is the analysis of the responses given by Panel members to these
questions. For sake of clarity, we have used a heat-map like plot to graphically display the most
cited levels of strength of interconnectedness between components.
Note that the term “interconnectedness” is divided into five levels of strength, each
represented by a color in the heat-map plot: i) dark blue for “Very Strong” interconnectedness,
ii) light blue for “Strong”, iii) pink for “Weak”, iv) red for “Very Weak”, and v) white for “Zero”
interconnectedness. For example, interconnectedness between the Biz-demand constraint and
the Safety and security value repository is most frequently labelled by Panel members as “Very
Weak”, thus this interconnection appears in the heat map plot in red color.
D.1 Interconnectedness between constraints and value repositories
The analysis of responses to Q1 shows a wide dispersion in the strength of most cited
interconnections existing between constraints and value repositories. This is visually apparent
by the fact that no color ―representing the strength of interconnectedness― dominates the
heat-map plot (Fig.15).
Some constraints such as Fuel cost, Capital intensity and Labor costs, although were
labelled key by Panel members in previous rounds, are now labelled as weakly interconnected
with value repositories, exception made for some value repositories ―i.e. Process and cost
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
182
optimization, Capacity management― for which strength of interconnectedness is mainly
labelled as strong or very strong.
(*) Z: Zero, VW: Very Weak, W: Weak, S: Strong, VS: Very Strong
Figure 15. Strength of Interconnectedness: Constraints-Value Repositories (Round 3)
Source: own elaboration
This observation contrasts with the (mainly) “Strong” interconnectedness shown by
other constraints, such as Commoditized product offering and Competition from other airlines.
At the same time, Slot availability and Excess capacity are the constraints with the weakest
interconnectedness of all.
The analysis from the value repositories side shows that only Process and cost
optimization has mainly “Very Strong” interconnectedness with constraints. On the contrary,
Safety and security is the value repository with the weakest interconnectedness of all.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
183
D.2 Interconnectedness among value repositories
The most frequently labelled strength of interconnectedness among value repositories
is summarized in the heat-map plot presented in Fig.16. The visual observation of the plot
suggests an almost total lack of red color, or “Very Weak” interconnections, with the exception
of the interconnectedness between Safety and Security and Revenue Management. This
contrasts with the fact that dark blue (“Very Strong”) and light blue (“Strong”) interconnections
are the predominant in the plot.
(*) Z: Zero, VW: Very Weak, W: Weak, S: Strong, VS: Very Strong
Figure 16. Strength of Interconnectedness: Value Repositories (Round 3)
Source: own elaboration
Some value repositories are particularly worth noting for their high strength of
interconnectedness. This is particularly the case of the Customer Experience and Customer-
centric proposition value repositories, which mainly present “Very Strong” interconnectedness
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
184
with other value repositories. Others, such as Distribution Management, Process and cost
optimization, and People and talent, also stand out for its “Strong” level of interconnectedness.
D.3 Interconnectedness between value repositories and Operating Margin
Panel members’ responses to the question of interconnectedness between value
repositories and Operating Margin mostly reflect either a “Very Strong” (VS) or “Strong” (S)
interconnectedness. This can be visually evidenced by observing Fig.17, where interconnections
of green (“Strong”) and blue color (“Very Strong”) dominate.
Figure 17. Strength of Interconnectedness: Value Repositories-Op. Margin (Round 3)
Source: own elaboration
It is worth highlighting the “Very strong” interconnectedness that Panel members assign
to value repositories such as Revenue Management, Network and Management/Leadership, not
Z: Zero
VW: Very Weak
W: Weak
S: Strong
VS: Very Strong
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
185
to mention the “Strong” interconnectedness given to Innovation, Alliances, and Distribution
Management, to name just a few examples.
E. Delphi round 4
Upon completion of Round 3, a report containing an individual benchmarking of the
responses given by each member of the Experts’ Panel with respect to the aggregated responses
of the Panel was drafted and handed out to the members (see example in Appendix B). The
report allowed Panel members to know the differences existing between their responses and
those of the Panel, so that they could be taken into account in Round 4 questionnaire.
Specifically, Round 4 questionnaire asked Panel members to reach consensus on the
interconnectedness among components, in addition to their strength and sign. These were the
questions made to Panel members:
Q1: Set the final strength and sign of the links between the 10 consensus value
constraints and the 15 consensus value repositories.
Q2: Set the final strength and sign of the links between the 15 consensus value
repositories.
Q3: Set the final strength and sign of the links between the 15 consensus value
repositories and airline's operating margin.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
186
E.1 Interconnectedness between constraints and value repositories
The comparative analysis of Panel members’ responses on the interconnectedness
between constraints and value repositories in rounds 3 and 4, leads us to conclude the
following:
There is a decrease in the interconnectedness mainly labelled as “Very Weak”; this
is visually reflected by the lower number of red interconnections shown in the heat-
map plot (Fig.18).
The interconnectedness labelled as “Zero” moderately increases from Round 3 to
Round 4, therefore many of the interconnections previously identified seem to lose
momentum.
The interconnectedness labelled as “Strong” and “Very Strong” by Panel members in
Round 3 practically keeps the same level of strength in Round 4.
By comparing the results gathered in rounds 3 and 4, we may conclude that Panel
members have fundamentally reached a broad consensus on the strength of the
interconnections labelled as “Strong” and “Very Strong”. Similarly, Panel members have opted
to reduce the strength of a good number of the interconnections mostly labelled as “Weak “ and
“Very Weak” in Round 3, changing them to a “Zero” level of interconnectedness in Round 4. This
should be no surprise, as this is part of a typical “experts consensus” building process driven the
Delphi process.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
187
(*) Z: Zero, VW: Very Weak, W: Weak, S: Strong, VS: Very Strong
Figure 18. Strength of Interconnectedness: Constraints-Value Repositories (Round 4)
Source: own elaboration
E.2 Interconnectedness among value repositories
The analysis of the responses on the interconnectedness among value repositories in
Delphi Round 4 reflects noteworthy changes with respect to Round 3. Most significant examples
of such changes is the disappearance of the interconnections mostly labelled as “Very Strong”
(dark blue color), whereas the predominant strength now becomes “Strong” (pink color), as we
can see in Fig.19.
Also worth noting is the growth in the number of (mainly) “Zero” (non-existent)
interconnections among value repositories with respect to Round 3, in addition to the increase
in the interconnections mainly labelled as “Very Weak”.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
188
(*) Z: Zero, VW: Very Weak, W: Weak, S: Strong, VS: Very Strong
Figure 19. Strength of Interconnectedness: Value Repositories (Round 4)
Source: own elaboration
E.3 Interconnectedness between value repositories and Operating Margin
Round 4 responses on the interconnectedness between value repositories and
Operating Margin mostly reveal an increase in the strength of these interconnections with
respect to Round 3.
As we can observe in Fig.20, the proportion of mostly labelled “Very Strong” (VS)
interconnections has increased in almost all value repositories, the only exceptions being the
interconnections of Alliances, which maintains the same strength as in Round 3, and Safety and
Security, which is now mostly labelled as a “non-existent” by Panel members.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
189
(*) Z: Zero, VW: Very Weak, W: Weak, S: Strong, VS: Very Strong
Figure 20. Strength of Interconnectedness: Value Repositories-Op. Margin (Round 4)
Source: own elaboration
5.3.2 Mapping the firm-system and Visualizing complexity
Once data on the key components of airlines and its interconnectedness has been
captured, it is time to map the firm-system and visualize the firm’s network complexity. For sake
of brevity, we have merged both the mapping and visualizing stages together (see Chapter 4),
which would otherwise be separated stages were this field research a detailed firm-specific
application of the CBVF methodology.
Key for our purposes at this point is to gather a thorough understanding of the airlines
value system dynamics. To achieve this goal, the image of a network is a natural one to use,
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
190
especially as we gather knowledge to better understand the behavior of the components of the
firm and their interconnectedness (Kolaczyk 2009). Notwithstanding the foregoing, the term
“network” is one used ambiguously and in a variety of ways ―e.g. as a graph, as a collection of
interconnected things. In this field research we use the term “network” in its most general form
to refer to a graph representing a network, or simply a “network graph”.
Among the many methods provided by modern “state-of-the-art” network analysis, the
one used by the author has to do with descriptive analysis of data. This approach primarily
involves the visualization and numerical characterization of the airlines’ value creation network.
Specifically, the author constructs a visual summary of the airlines’ value creation network,
which in turn enables us to combine the data collected in the Delphi process into one single
digraph from where we can compute diverse topological complexity methods.
Under this approach, a graph made of three different types of vertices ―representing
the constraints, value repositories, and Operating Margin― and edges ―representing the
interconnection between pairs of vertices― is used to represent the firm’s network.
A. Visualizing the Airlines’ Value Creation Network (AVCN)
In this section we address the task of visualizing the airlines’ value creation network
(AVCN). In other words, we focus on drawing a graph that can help us visualize the AVCN and
appraise key topological complexity methods.
We start by assessing different ways to lay out the AVCN graph, followed by some ways
to decorate such layout using a combination of mathematics, human aesthetics, and algorithms.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
191
Fig.21 offers a first approach to the graph of the AVCN, considering different levels of strength
of interconnectedness between constraints, value repositories and Operating Margin.
A first impression that we form when we observe the AVCN graph at the top of Fig.21 is
that of a heavily connected graph, where everything seems highly connected to everything else.
This is certainly a “clogged” graph that makes it visually difficult for an observer’s eye to discern
between links. It is needless to say that the AVCN graph at the top does not help much if we
seek clarity, nor does it allow us to infer valuable conclusions from a visual standpoint, other
than the AVCN is inherently complex.
To appraise key topological complexity methods and draw meaningful conclusions at
this stage of the CBVF methodology, such as what are the key nodes or key interconnections in
the AVCN, one possible solution is to discriminate the number of interconnections by strength.
Following this method, the graphs in the middle and at the bottom of Fig.21 display only the
links of higher strength, with the graph at the bottom showing only those links labelled as “Very
Strong” by Panel members. As we note, this latter graph is made up of the same 26 nodes as the
graph at the top, but with only 77 of the 319 links; likewise the network degree distribution
―which gives us an idea of the distribution of connections in the network― noticeably
decreases from 29.99 to 5.92. Now the AVCN appears much more tidy and clear to our eyes.
This progressive reduction in the number of links in the AVCN graph enables us to infer
the following valuable observations:
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
192
Figure 21. Airlines’ Value Creation Network (AVCN)
Source: own elaboration, based on Pablo-Martí, Muñoz-Yebra et al. 2014
Type: directed-weighted
Layout: Fruchterman-Reingold
Colors: Red-circle: constraints,
Yellow-triangle: value repositories,
Orange-square: Operating Margin
Nodes: 26
Edges: 390
15 self-loops
Degree: 29.99
All existing interconnections are
displayed
Type: directed-weighted
Layout: Fruchterman-Reingold
Colors: Red-circle: constraints,
Yellow-triangle: value repositories,
Orange-square: Operating Margin
Nodes: 26
Edges: 238
15 self-loops
Degree: 18.31
Only Strong & Very Strong
interconnections are displayed
Type: directed-weighted
Layout: Fruchterman-Reingold
Colors: Red-circle: constraints,
Yellow-triangle: value repositories,
Orange-square: Operating Margin
Nodes: 26
Edges: 77
15 self-loops
2 isolated nodes
Degree: 5.92
Only Very Strong interconnections
are displayed
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
193
Some value repositories are more critical than others when it comes to value
creation ―e.g. Revenue management, Capacity management, Customer experience,
Customer-centric proposition, Innovation.
Some value repositories are much more influenced by constraints than others ―e.g.
Process and cost optimization, Brand, Revenue management, Capacity
management.
Some constraints are much less relevant to value creation than we might have
initially thought from experience ―e.g. Slots availability and Capital intensity.
Only nine value repositories significantly affect Operating Margin.
Using a network scaling method is of great help at this point to address the above
observations in more detail and gather an in-depth knowledge on the underlying structural
architecture of the AVCN and its building process.
Specifically, the author used a Pathfinder Network Scaling method. This method relies
on the so-called triangle inequality to eliminate redundant or counter-intuitive links. Given two
links or paths in a network that connect two nodes, the link/path preserved is the one with a
greater weight defined via the Minkowski metric. It is assumed that the link/path with the
greater weight better captures the interrelationship between the two nodes and that the
alternative link/path with less weight is redundant or even counter-intuitive and should be
pruned from the network (Indiana 2005).
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
194
Figure 22. AVCN Pathfinder Network Scaling
Source: own elaboration
A
B
C
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
195
Two parameters and influence the topology of a pathfinder network. The
influences the weight of a path based on the Minkowski metric. The
defines the number of links in alternative paths, up to which the triangle inequality
must be maintained. A network of nodes can have a maximum path length of .
With the triangle inequality is maintained throughout the entire network
(Schvaneveldt, 1990).
Fig.22 contains three graphs (A, B and C) simulating the transition of the AVCN through
three different stages of its building process. Each graph incrementally assigns higher distances
or dissimilarities among the set of links and the set of nodes upon their relative positions. The
graphs so created enable us to explain the underlying organization of the AVCN in a highly
schematic way, as well as to visually determine its key components and interconnectedness with
Operating Margin (OM). Furthermore, we are able to graphically visualize the clustering
processes growing around the airlines’ Operating Margin in the form of two incipient groups of
value repositories marked by two red circles (C).
The analysis of the AVCN graph can be taken one step further by using Hierarchical
Clustering (HC) and a tree-like dendrogram (Figs. 23 and 24). The HC algorithm produces, as the
name indicates, an entire hierarchy of nested partitions in the structure of value repositories,
comprising the following clusters when the strength of the interconnections is “Strong” or “Very
Strong”:
Cluster 1 (Red), contains only two value repositories: Information management and
Process and cost optimization.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
196
Cluster 2 (Green), contains six value repositories: Network, Revenue management,
Capacity management, Alliances, Distribution management, and Customer-centric
proposition.
Cluster 3 (Blue), contains seven value repositories: People and talent, Management-
Leadership, Safety and security, Corporate culture, Innovation, Brand, and Customer
experience.
Figure 23. Clusters in the AVCN
Source: own elaboration
The HC algorithm takes the distance information ―dissimilarity― between the value
repositories, and links the pairs of them that are close together. This process is repeated linking
these newly formed clusters to each other and to other value repositories in order to create
bigger clusters, until all of them are linked together in a hierarchical tree or tree-like
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
197
dendrogram (Fig.24). The height of the links indicates the distance between the value
repositories.
Figure 24. AVCN Tree-like Dendrogram
Source: own elaboration
B. Descriptive methods of AVCN complexity
A summary of some key topological methods available to comprehend complexity of the
AVCN is presented in Appendix F, subsection 2.4. These methods mostly derive from areas
outside mainstream statistics and have traditionally been used for descriptive purposes
(Kolaczyk, Csárdi 2014).
An overwhelming proportion of these methods are graph-theoretic in nature, and have
their origins in mathematics and computer science. More recently, the field of physics has also
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
198
been an important contributor, with many methods often motivated by analogues in statistical
mechanics.
The topological network complexity methods presented in Appendix F include vertex
and edge characteristics, measures of network cohesion, and assortativity and mixing.
C. What value repositories are most important?
To determine the most important or influential value repositories within the AVCN, we
use a network centrality measure known as Eigenvector centrality. This measure is based on the
notion of “status” or “rank” of the value repositories specified, and seeks to capture the idea
that the more central the neighbors of a value repository are, the more central that value
repository itself becomes (Kolaczyk, Csárdi 2014). The Eigenvector centrality is therefore an
approximate measure of the importance of each value repository within the AVCN.
As we can see in Table 9, upon computing Eigenvector centrality for the AVCN we can
conclude that Customer-centric proposition is the most influential/important value repository of
all. Based on this first outcome, we continue to assign a maximum score of one to the
importance of Customer-centric proposition and compute the importance of each subsequent
value repository with respect to the former.
Value Repository
Importance
Customer-centric proposition
100
Brand
98
Customer experience
95
Revenue management
94
Network
88
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
199
Value Repository
Importance
Process and cost optimization
87
Distribution management
86
Corporate culture
80
Alliances
79
Management (Leadership)
79
Capacity management
78
People and talent
78
Information management
73
Innovation
70
Safety and security
50
Table 9. Most Important/Influential Value Repositories in the AVCN
Source: own elaboration
It is worth noting that customer-related value repositories ―e.g. Brand, Customer
experience― are among the most important/relevant value repositories in the AVCN, whereas
Safety and security is the least important/relevant among the top fifteen value repositories.
D. What affects most the Operating Margin?
Up to this point we have identified nine value repositories that affect airlines’ Operating
Margin. Now we are interested in finding out further details on what those value repositories
are, and what the strength of the interconnections between each value repository and
Operating Margin is.
We resort to the AVCN graph to visually determine the interconnections having an
impact on the airlines’ Operating Margin. As we can see in the graph on the left in Fig.25, all top
fifteen value repositories identified by Panel members have some kind of interconnection with
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
200
Operating Margin. What varies from one value repository to another is the strength of
interconnectedness.
In contrast, the graph on the right shows that only five value repositories have a “Very
Strong” interconnectedness with Operating Margin: Network, Capacity management, Revenue
management, Management/Leadership, and People and talent; whereas those with a “Strong”
interconnectedness are four: Innovation, Process and cost optimization, Customer experience,
and Customer-centric proposition.
Network (NW), Revenue management (RV), People and talent (PE), Management/Leadership (MM), Corporate culture (CU), Capacity m anagement
(CM), Customer experience (EX), Innovation (IV), Brand (BD), Alliances (AL), Safety and security (SF), Distribution management (DI), Process an d cost
optimization (OP), Information management (IN), Customer-centric proposition (CC), Operating Margin (OM)
Left: All Value Repositories connecting with Operating Margin. Right: Value Repositories with “Strong” (black arrows)
and “Very Strong” (green arrows) interconnectedness.
Figure 25. Interconnectedness of Value Repositories and Operating Margin
Source: own elaboration
5.3.3 Modeling and simulation
Previous stages in this thesis field research have provided us with the data needed to
build the airlines’ value creation network (AVCN), to map the interconnectedness between the
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
201
different AVCN components, and between these and Operating Margin. Now according to the
CBVF methodology (Chapter 4), it is time to address the question on how to convert the AVCN
into a practical model that enables airlines’ managers simulate future scenarios and anticipate
the impact on performance.
The technique chosen by the author for modeling and simulation consists of a Fuzzy
Cognitive Map (FCM). The FCM is a soft-computing technique that enables researchers and
practitioners to model the properties of the AVCN from expert knowledge and determine
possible future states and instabilities of the firm’s network (see Chapter 4, Section 4.3.7, for a
detailed description on FCMs). The FCM is not the only technique available to approach the
CBVF, there are of other methods and tools that could have been used in this field research,
such as those described in Chapter 4. However, the scope and method of our field research
made this technique specially suitable for the goals pursued.
In particular, FCMs are most useful when other more refined quantitative methods fail.
This usually occurs in broad knowledge domains with only partial experts, in situations with little
or no relevant historic data, and in cases where most information is qualitative and fuzzy. FCM
models use a mix of qualitative and quantitative approaches, consider multivariate interactions
leading to nonlinearities, and aim to make implicit assumptions (or mental models) explicit
(Jetter, Kok 2014). These properties of FCMs meet altogether our field research modeling and
simulation needs.
Once the AVCN-FCM model has been formulated, the subsequent simulation tasks are
performed by studying how uncertainty in the outcome variable ―Operating Margin― can be
attributed to different changes in the source of inputs ―i.e. value repositories and constraints.
This simulation technique, known as Sensitivity Analysis (SA), aims at analyzing the behavior of
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
202
the firm response locally around the interconnectedness of the AVCN components and
Operating Margin.
The importance and usefulness of SA is widely recognized in the literature (Saltelli 2002,
Cacuci, Ionescu-Bujor et al. 2005, Carrillo-Hermosilla 2015), where we can find a good number
of reviews and applications of SA in the various areas of applied economics ―i.e. decision
making, communication, increased understanding of systems, and model development, among
others. For some authors a sensitivity analysis should be a prerequisite for model building and
“an integral part of any solution methodology” (Pannell 1997), without which the status of a
solution could not be understood. In this field research, the SA is a method used for the
corroboration, quality assurance, and demonstrability of our AVCN-FCM based analysis (Saltelli
2002).
It is worth noting that the ultimate goal of the FCM is not to create a ‘‘true’’ model of
the AVCN, but a useful and formalized description of the perception of a group of experts in the
airline industry. A benchmark that gauges the validity of the FCM should thereafter assess
whether the FCM adequately describes what the Panel members know about value creation in
airlines. This in turn would require Panel members to take an active role in practical model
testing.
A. The AVCN-FCM model
Construction of our AVCN-FCM model consists of a multi-step process that captures
knowledge from experts in the form of a network or map, formally describes the AVCN as an
adjacency matrix, and applies neural network computation to refine the model and analyze the
results.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
203
For the purpose of this field research, the FCM model construction framework consisted
of three steps (Jetter, Kok 2014):
1. Knowledge capture (Step 1). This process step included all the elicitation activities
that led to formulate one individual weighted causal cognitive map describing the
AVCN for each Panel member.
2. Detailed design of the FCM inference model (Step 2). Included the aggregation of
the individual causal cognitive maps of Step 1 into one single relational adjacency
matrix, and the design and implementation of the FCM inference engine.
3. Simulation and interpretation of results (Step 3). This was accomplished by choosing
the input vectors and carrying out a SA on the basis of real-life airlines scenarios.
A.1 Step 1: Knowledge capture
As seen in previous sections, the Delphi process is the technique used in this field
research to capture knowledge from Panel members. Therefore, the Delphi process itself
encompasses all the elicitation activities needed to produce the members’ individual cognitive
maps, prior to the construction of the final (aggregated) AVCN-FCM model.
Furthermore, as far as the analysis of the data captured through the Delphi process is
complemented with the mapping of the firm-system and visualization of complexity activities
(Section 5.3.2), we end up having a precise idea on how the AVCN-FCM looks like and what the
critical structural and relational parameters for the construction of the FCM model must be.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
204
A.2 Step 2: Design of FCM model
In Step 2 we mathematically aggregate the individual cognitive maps gathered from the
Panel members into a single integrated cognitive map. This is accomplished by first translating
each individual map into square adjacency matrices of the same size. This operation results in a
new matrix, the entries of which are the average of the weights of the interconnections assigned
by the members.
Weights in the matrix are based on a Likert scale with five different levels of strength,
expressed as: zero (0), very weak (0.25), weak (0.5), strong (0.75), and very strong (1). Note that
the numerical values of the weights indicate the degree of influence between component (or
concept) and component (or concept) .
Also incorporated into the adjacency matrix are the signs of the causal relations
between the components of the AVCN, ranging from positive (+1) to negative (-1). A positive
causality between component and component means that an increase of the value of
component will cause an increase in the value of component , likewise a decrease of the
value of component will cause a decrease in the value of component . Negative causality
between components means that an increase in the first component will cause a decrease in the
value of the second component, or that a decrease of the first component will cause the
increase of the second. When no relationship exists between two components, the weight
assigned by Panel members is zero.
The aggregated AVCN-FCM adjacency matrix used in our field research is shown in Table
10.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
205
Nodes # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
Regulation 1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.75 0.5 0.00 0.63 0.00 0.75 0.00 -0.75 0.00 -0.75 1.00 -0.75 0.75 0.25 - 0.5 0.00
Fuel 2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.75 0.63 0.00 0.00 0.00 1.00 -0.13 0.75 0.00 0.00 0.00 0.13 0.75 0.13 0.5 0.00
Competition 3 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -1.00 -1.00 0.75 0.75 0.75 1.00 0.75 0.88 0.88 1.00 0.00 0.88 0.75 0.75 0.75 0.00
Commoditization 4 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.5 0.5 0.13 0.75 0.63 0.75 -0.75 0.75 0.88 0.75 0.00 0.75 0.75 0.75 0.75 0.00
Unions 5 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.5 0.38 -0.75 -0.75 -0.75 -0.13 0.13 - 0.5 0.25 0.25 0.38 0.00 -1.00 0.13 0.00 0.00
Labor 6 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.5 0.5 0.75 0.75 0.75 -0.38 -0.25 -0.38 0.00 0.5 0.00 0.00 -0.88 0.00 0.00 0.00
Slots 7 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.75 0.13 0.13 0.25 0.13 0.75 0.25 0.25 0.00 0.63 0.00 0.25 0.25 0.00 0.25 0.00
ExCapacity 8 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.75 -1.00 0.00 0.38 0.25 -1.00 0.38 0.5 0.13 0.75 0.00 0.63 0.75 0.00 0.13 0.00
Capital 9 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.75 0.00 0.00 0.63 0.13 0.75 0.13 0.75 0.00 0.00 0.00 0.00 0.75 0.00 0.00 0.00
Biz-demand 10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.75 1.00 0.00 0.13 0.38 1.00 0.75 0.75 0.75 0.75 0.00 0.75 0.25 0.38 0.75 0.00
Network 11 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.5 0.00 0.25 1.00 0.75 0.5 0.75 1.00 0.00 0.75 0.75 0.5 0.75 1.00
Revenue 12 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.5 0.5 0.25 1.00 0.00 0.00 0.00 0.75 0.00 1.00 0.75 0.75 0.75 1.00
People 13 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.75 0.75 1.00 1.00 1.00 0.75 1.00 0.75 1.00 0.5 0.75 0.00 0.75 0.75 1.00 1.00
Management 14 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.75 0.75 1.00 1.00 1.00 0.75 0.75 1.00 1.00 0.75 1.00 0.75 0.75 0.75 1.00 1.00
Culture 15 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.75 0.75 1.00 1.00 1.00 0.00 1.00 1.00 1.00 0.00 0.75 0.25 0.75 0.5 1.00 0.75
Capacity 16 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.25 0.00 0.00 1.00 0.75 0.00 0.5 0.00 0.25 0.75 0.75 0.00 0.00 1.00
Experience 17 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.5 0.5 0.75 0.75 0.75 0.00 1.00 0.75 1.00 0.75 0.00 0.75 -0.75 0.75 1.00 0.88
Innovation 18 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.75 0.75 0.75 1.00 1.00 0.5 1.00 1.00 1.00 0.5 0.75 0.75 0.75 0.75 1.00 0.88
Brand 19 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.5 0.75 0.75 0.75 0.5 1.00 0.75 1.00 0.75 0.5 0.75 0.00 0.00 0.75 0.75
Alliances 20 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.75 0.00 0.00 0.5 0.75 0.5 0.5 0.75 1.00 0.00 0.75 0.5 0.5 0.75 0.75
Safety 21 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.25 0.25 0.00 0.75 0.75 0.5 0.00 0.00 0.75 0.00 1.00 0.25 0.75 0.00 0.5 0.5
Distribution 22 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.75 0.75 0.25 0.5 0.25 0.5 0.75 0.5 0.5 0.75 0.00 1.00 0.75 0.75 0.75 0.75
Optimization 23 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.5 0.75 1.00 0.75 0.75 0.75 0.75 0.00 0.75 0.75 0.75 0.75 1.00 0.75 0.75 0.88
Information 24 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.75 0.75 0.5 0.75 0.75 0.5 0.75 0.75 0.75 0.75 0.75 0.75 0.75 1.00 0.75 0.75
Customer-centric 25 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.75 0.75 0.75 0.75 0.75 0.75 1.00 0.75 1.00 0.75 0.00 0.75 0.75 0.75 1.00 0.88
OpMargin 26 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Table 10. AVCN-FCM Adjacency Matrix
Source: own elaboration
A.3 Step 3: Simulation and interpretation of results
Step 3 is the stage where the AVCN-FCM model is initialized; in other words, this is the
stage where we set the initial vector value that feeds the model. Once the initial vector has been
fed in, the AVCN-FCM model can start performing simulations in a series of iterations.
Input vectors can be chosen to reflect any specific constraint setting, value creation
scenario, or policy choice made by the firm. For the purpose of this research, the author used a
random input vector as a Base scenario, against which different variant scenarios were later
compared.
At each running step, the value of the components is recalculated according to equation
(1).
(1)
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
206
Where is the value of component at time , is the value of
component at time , is the weight of the interconnection between component
and component , and is the sigmoid function:
(2)
In FCM terminology, a running step is defined as the time step during which the values
of the components are calculated. As seen in (1), the value of each component is defined by
taking all the causal linkage weights pointing to this component and multiplying each weight by
the value of the component that causes the linkage. Then, the sigmoid function is applied and
an outcome results in the range .
A Hebbian-based learning algorithm, known as Differential Hebbian Learning (DHL) is
used for learning the adjacency matrix. The DHL algorithm assumes that, if two concepts on the
opposite side of an edge change simultaneously, then the weight of that edge is increased. The
difference for every concept is computed as Δ, if the activation
of the concept changes. If , then the weight between concepts (cause) and (effect) is
also changed, therefore . If , then the
corresponding weight does not change, therefore .
During DHL learning, the values of weights are iteratively updated until the desired
structure is found. The weights of outgoing edges for each concept in the adjacency matrix are
modified only when the corresponding concept value changes (Papageorgiou 2012).
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
207
B. Scenario simulation
Scenario simulation using the AVCN-FCM model can take place in a variety of ways. For
example, we might use a structural assessment simulation to compare the AVCN structure with
that of a real-life airline, or we might want to conduct an extreme conditions test, namely to test
if extreme constraints and/or value repositories values lead to unmanageable airline behavior.
Yet, despite how revealing these techniques might be, they would involve a significant amount
of airline-specific knowledge, the capture of which lies outside of the scope and method of this
field research.
As we have previously explained, the simulation technique chosen in this field research
is based on Sensitivity Analysis (SA). SA specifically examines firm’s components influence on
Operating Margin upon simultaneously changing input parameters. Note importantly that SA is
initiated by feeding a random input vector into equation (1), thus obtaining an outcome state
vector of reference. A random initialization of the model is required as no absolute valuation of
the airlines’ components was practicable within the scope of the field research.
Table 11 summarizes the three exemplary scenarios simulated in this stage:
Scenario
#
Goal
Related AVCN
Components/Constraints
1
Assess the impact of a tougher
business environment
Regulation
Fuel cost
Competition from other airlines
Labor cost
Business travel demand
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
208
Scenario
#
Goal
Related AVCN
Components/Constraints
2
Assess the impact of better
customer-oriented processes
Process and cost optimization
Customer-centric proposition
Innovation
Information management
Customer experience
3
Assess the impact of more
productive asset management
Network
Distribution management
Capacity management
Table 11. AVCN-FCM Simulation Scenarios
Source: own elaboration
B.1 Base scenario
A Base scenario is formulated on the basis of a random input vector, which is fed into
the AVCN-FCM model to generate an outcome state vector of reference. From this Base
scenario, new variant scenarios (Table 11) are formulated by simply varying the values of the
random input vector. Once these variant input vectors are fed themselves into the model, the
deviations between the variant scenario outcomes and the Base scenario outcome are
measured and interpreted.
The inference outcome for the Base scenario is shown in Fig.26. As we can see, after
performing the iterative inference process for fifty steps, the model reaches an equilibrium
point at which the AVCN state vector becomes stable and does not change anymore.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
209
Figure 26. Base Scenario Inference Outcome
Source: own elaboration
Table 12 contains the outcome state vector resulting from the Base scenario, to which
we will next compare our three scenarios variants.
Table 12. Base Scenario Outcome State Vector
Source: own elaboration
Operating Margin
Op. Margin
0.83
Base scenario- AVCN Outcome State Vector
Network
0.64
Capacity
0.66
Safety
0.64
Revenue
0.69
Experience
0.74
Distribution
0.68
People
0.72
Innovation
0.68
Optimization
0.61
Management
0.69
Brand
0.73
Information
0.53
Culture
0.72
Alliances
0.60
Customer-centric
0.70
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
210
B.2 Scenario #1: tougher business environment
This simulation was conducted to assess the impact of a tougher business environment
on the AVCN components and Operating Margin. In particular, a tougher business environment
was defined as one with a more strict regulatory framework, higher resources costs, higher
competition from other airlines, and lower business travel demand.
For the purpose of the model inference, we used as input values those contained in
Table 13. Note that all values of the input vector not contained in this table remain the same as
in the Base scenario.
AVCN Component
Input value
Regulation
-0.9
Fuel
-0.9
Competition from other airlines
-0.9
Labor
-0.9
Business travel demand
-0.9
Table 13. Scenario #1 Input Vector
Source: own elaboration
The inference outcome for Scenario #1 is displayed in Fig.27. The model reaches
equilibrium after performing a number of iterations, which means that outcome stability is
achieved.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
211
Figure 27. Scenario #1 Inference Outcome
Source: own elaboration
Table 14 below contains the values of the final state vector for Scenario #1 and shows
the differences with respect to the Base scenario outcome.
Scenario #1- AVCN Outcome State Vector & Differences with Base scenario
Network
0.56
-12.3%
Capacity
0.46
-30.5 %
Safety
0.57
-11.8%
Revenue
0.60
-12.7%
Experience
0.72
-2.4%
Distribution
0.55
-18.6%
People
0.70
-8.5%
Innovation
0.61
-10.5%
Optimization
0.54
-10.0%
Management
0.63
-9.4%
Brand
0.61
-17.0%
Information
0.53
0.0%
Culture
0.66
-8.5%
Alliances
0.47
-20.3%
Customer-
centric
0.64
-9.3%
Operating Margin
Op. Margin
0.80
-3.0%
Table 14. Scenario #1 Outcome State Vector
Source: own elaboration
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
212
We can see that as result of the simulation, the Operating Margin is expected to
decrease before a tougher business environment in accordance with airlines’ business reality.
Also worth noting is the drop in the Capacity management value repository, which is the most
negatively affected of all firm’s components. This, in turn, significantly affects the revenue,
distribution and resource related value repositories and, very specifically, those related to
people ―most probably due to corporate layoffs.
Moreover, according to our simulation, a tougher business environment would lead to a
negative impact on the customer-centered value repositories ―e.g. Brand, Customer experience
and Customer-centric proposition―, as well as those dedicated to organizational improvement
―e.g. Innovation, Information management and Process and cost optimization. In summary, our
simulation of a tougher business environment would generate a shock on the viability of the
airline itself.
B.3 Scenario #2: implementation of better customer-oriented processes
Our next simulation was conducted to assess the impact of better customer-oriented
processes on the AVCN components and the Operating Margin. To perform this simulation, we
considered an increase of the input values of the following value repositories:
Process and cost optimization.
Innovation.
Customer experience.
Information management.
Customer-centric proposition.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
213
For the purpose of the model inference, these changes translated in the following input
values to the AVCN-FCM model, all other values of the input vector remaining the same as in the
Base scenario (Table 15).
AVCN Component
Input value
Process and cost optimization
0.9
Innovation
0.9
Customer experience
0.9
Information management
0.9
Customer-centric
0.9
Table 15. Scenario #2 Input Vector
Source: own elaboration
The inference outcome for Scenario #2 appears in Fig.28, where we can observe how
outcome stability is reached after a number of inference iterations.
Figure 28. Scenario #2 Inference Outcome
Source: own elaboration
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
214
The resulting outcome state vector from Scenario #2, together with the deviations from
the Base scenario, is shown in Table 16.
Scenario #2- AVCN Outcome State Vector & Differences with Base scenario
Network
0.64
0.1%
Capacity
0.66
0.1%
Safety
0.64
0.6%
Revenue
0.70
2.2%
Experience
0.90
NA
Distribution
0.68
1.2%
People
0.75
4.3%
Innovation
0.90
NA
Optimization
0.90
NA
Management
0.72
4.0%
Brand
0.75
2.4%
Information
0.90
NA
Culture
0.75
4.3%
Alliances
0.59
0.0%
Customer-
centric
0.90
NA
Operating Margin
Op. Margin
0.86
4.0%
Table 16. Scenario #2 Outcome State Vector
Source: own elaboration
The outcomes derived from the implementation of better customer-oriented processes
indicate that a significant increase on the airline’s Operating Margin is to be expected.
Moreover, this scenario would affect positively people-related value repositories ―e.g. People
and talent, Management/Leadership and Corporate culture―, in line with what real-life airline
managers would expect. This would help explain the key role of people in firms carrying out
customer-oriented policies.
Last but not least, Scenario #2 would also affect positively the perception of the brand
by consumers, as well as the Revenue management and Distribution management value
repositories, most probably due to the increased business activity generated by more loyal
customers.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
215
B.4 Scenario #3: more productive asset management
Our third simulation was conducted to assess the impact of higher values in those value
repositories associated with airlines’ asset management, namely Network, Revenue
management, Capacity management, and Distribution management.
For the purpose of the model inference, we used as input values those contained in
Table 17, all remaining values of the input vector remaining the same as in the Base scenario.
AVCN Component
Input value
Network
0.9
Revenue management
0.9
Capacity management
0.9
Distribution management
0.9
Table 17. Scenario #3 Input Vector
Source: own elaboration
The analysis of the simulation inference outcome (Fig.29), and of the final AVCN state
vector (Table 18), shows the positive impact that this simulation has on the airline’s Operating
Margin. This outcome is therefore in line with what airline’s managers would expect in real-life
business. Also note that stability of the outcome is reached after a number of iterations, as can
be observed in Fig.29.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
216
Figure 29. Scenario #3 Inference Outcome
Source: own elaboration
Scenario #3- AVCN Outcome State Vector & Differences with Base scenario
Network
0.90
NA
Capacity
0.90
NA
Safety
0.64
0.0%
Revenue
0.90
NA
Experience
0.75
1.9%
Distribution
0.90
NA
People
0.72
0.1%
Innovation
0.68
0.0%
Optimization
0.61
0.0%
Management
0.69
0.0%
Brand
0.73
0.0%
Information
0.53
0.0%
Culture
0.72
0.1%
Alliances
0.61
1.7%
Customer-
centric
0.70
0.1%
Operating Margin
Op. Margin
0.85
2.9%
Table 18. Scenario #3 Outcome State Vector
Source: own elaboration
It is worth noting that the impact of Scenario #3 on most value repositories is rather
limited if compared to the effects of Scenario #2. As we can see in Table 18, Scenario #3 final
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
217
state vector shows many value repositories not being really affected by the changes in the
values of the input vector, which is otherwise consistent with airlines’ business reality.
5.4 Discussion and implications
The empirical findings from our field research show certain trends that are useful for
understanding complexity in airlines and lead to the generation of 10 tentative propositions
that summarize some of the most important aspects of the CBVF theoretical narrative:
1. Some value repositories are more critical than others when it comes to value
creation; some are much more influenced by constraints than others; and some
constraints are much less relevant to value creation than we might have initially
thought.
2. The strongest impact on value repositories is expected from only four different
constraints ―e.g. Competition from other airlines, Business travel demand, and
Commoditized product offering.
3. The value repository most affected by constraints is Process and cost optimization.
On the contrary, Safety and security is the value repository least affected by
constraints.
4. Interconnectedness among value repositories is mostly built around strong
interconnections, with no interconnections mainly labelled as “very strong” by Panel
members.
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
218
5. Interconnectedness between value repositories and Operating Margin is mostly
revealed as “strong” and “very strong”. The only exceptions being the
interconnections of the Alliances and Safety and Security value repositories.
6. The strongest impact on airlines’ Operating Margin comes from only five value
repositories, with four more value repositories having a moderate impact.
7. Customer-centric proposition is the most influential value repository of all, with
other customer-centered value repositories, such as Brand and Customer
experience, closely following in importance.
8. AVCN topological complexity measures feature the airlines’ network context as a
multi-layered clustered network embedded in a highly dimensional space, in
accordance with CBVF features.
9. The AVCN is built around three closely interconnected clusters, which themselves
are linked to create bigger clusters and form a hierarchical nested tree structure.
10. Outcomes from the AVCN-FCM model simulations are consistent with airlines’
business reality and they have been proven to have significant management
implications.
Furthermore, the above propositions indicate the applicability of the main theoretical
constituents of the CBVF, as stated in Chapter 4. Decomposition of the firm’s architecture into a
network of unique, differentiated, and autonomous value repositories has not only been a
conceptualization accepted by airlines’ experts, but also a valid construct that matches
complexity of the firm.
Moreover, the high level of participation achieved from Panel members, the positive
feedback received from them, and the quality of the contributions made, endorse our
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
219
assumption that the CBVF approach allows for a flexible and realistic understanding of the firm
and of the interactions that shape its behavior. And the theoretical constituents of the CBVF,
such as emergence, nonlinearity, hierarchical structure, homeostasis, and
differentiation/specialization, have been proven fully consistent with the findings of our testing
exercise of the theory and method of the CBVF in the airline industry.
More particularly, the AVCN-FCM model seems well suited to study emergent behavior
and feedbacks, and to test unintended side effects of various firm policy interventions. The
various scenarios tested during our Sensitivity Analysis prove, for example, how firm’s value
repositories reflect a value creation-exchange duality, and differentiation by specialization, thus
confirming our original theoretical assumptions. This is also true for the homeostasis property,
which becomes apparent when airlines achieve relative performance stability ―as reflected by
Operating Margin― before significant shocks from the outer environment (see Scenario #1 in
Section 5.3.3).
Our testing exercise also reveals that the application of the CBVF methodological
framework can potentially offer a wide range of possibilities for better approaching complexity
of the firm; from extracting relevant data that can be used to model the firm’s network context,
to simulate real-life scenarios that support competitiveness analysis and decision-making.
The FCM-based model has shown that it can be a key method for scenario generation in
firms struggling to assess its own value creation policies and anticipate performance. The
resulting scenarios might be used as a standalone output, or as an input for the generation of
more detailed value creation roadmaps or corporate strategies. Furthermore, the AVCN-FCM
model might help create the basis for further discussion on complexity between C-level
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
220
executives, management teams and stakeholders, enabling them to contrast simulations against
opinions, or comparing them to already existing scenarios.
Nevertheless, this author is well aware of the limitations that this testing exercise has.
For example, according to frequent criticism found in the literature, some authors might point
out as a limitation the difficulty in generalizing the results to a wider population of firms due to
sample size (Schmidt, Lyytinen et al. 2001) or the author’s own agenda (Nambisan, Agarwal et
al. 1999). These authors would most probably recommend further study to refine and verify the
results, or to investigate related sets of research questions (Skulmoski, Hartman et al. 2007).
Certainly, a more careful selection of cases according to the CBVF theoretical
constituents and the use of a cross-case analysis could alleviate this issue (Helper 2000). Whilst
we may agree with this reasoning, the author designed this testing experiment trying to prevent
bias throughout the field research. Rather than concentrating on a single airline and a small
number of experts, he stayed focused on a broad and diversified group of experts from different
network airlines across the word, holding a wide range of positions and performing in different
functional areas. Notwithstanding the foregoing, it should be borne in mind that no
methodology is perfect ―even regressions have serious problems of generalizability and
subjectivity― and that further verification study of the CBVF theory and method might provide
new and rich results.
This author is also aware that the FCM-based model, formulated as part of the modeling
stage of the field research, is contingent on the experts participating, the questions made
throughout the Delphi process, and the method used to aggregate the data captured. Had other
experts taken part in the field research, or had the questions been different or been made in
another moment of time, the results obtained may have also been different. Consequently, not
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
221
only the structural specification of the firm is contingent on the method used and the moment
in time, but also the type and strength of the interactions reflected among firm’s components.
That said, the above considerations should not jeopardize the usefulness of the CBVF approach
to tackle complexity of the firm, nor should its outcomes be neglected. Quite the contrary, all
the potentially different representations of the firm that we could obtain under the CBVF shall
be considered robust enough so as to deem the CBVF a valid theoretical-practical approach.
Other authors might allude to limitations inherent to the Delphi process, which in turn
would have several design and implementation implications. These might include i) the
somewhat “subjective” consensus method used in the field research (von der Gracht 2012), ii)
the higher/lower relevance given to response rates, and the suitability of the profile of the
participants, iii) the consideration that the iterative characteristics of the Delphi process can
potentially enable investigator to mold opinions, iv) the restrictions posed by the author to the
number of constraints (10) and value repositories (15) that might have contributed to the
stream of this field research, so that important minority issues may have been missed due to
nonconformity of general opinion, and v) the loss of objectivity and researcher bias in analyzing
findings and generating questions (Balasubramanian, Agarwal 2013). Although this author
agrees that some of the aforementioned limitations associated to the Delphi process might
certainly need further attention in future research, he contends that the Delphi process is a key
research method that has proven to be valid for extracting knowledge, which would not
otherwise have been obtainable by other means.
Still other authors might remain skeptical of this field research fearing that it is not
replicable; in other words, that it is difficult for other researchers to replicate this study by
interviewing the same people. Notwithstanding, the author has attempted to facilitate
CHAPTER 5. TESTING THE THEORY AND METHOD: AIRLINES FIELD RESEARCH
222
replicability of this field research by providing all the data captured, the questionnaires, and “R”
code transcripts, so as to enable other researchers to replicate this research elsewhere.
Nevertheless, it is worth recalling that the CBVF method is not about presenting the subjective
experience of experts per se, but about abstracting it into a set of tentative theoretical
propositions. In the CBVF methodology this conceptual abstraction is achieved through constant
comparison ― i.e. constant iteration between data collection and analysis― until the stage of
theoretical verification is achieved.
CONCLUSIONS
223
CONCLUSIONS
This thesis is an attempt to establish the foundations of both a complexity-based theory
of the firm and a practical method to enable researchers and practitioners to approach
complexity systematically and with confidence, through a more realistic understanding of the
behavior of the firm. The final aim being to foster an open discussion in the social research
community on the need to embed complexity awareness in the theory of the firm, as well as to
test innovative research that contributes to accelerate the implementation of a CBVF paradigm.
It is in pursuit of the above objectives that we have built our key assumptions on top of
propositions that are familiar to conventional theories of the firm, from where we have made
further progress at the theoretical, as well as the methodological and empirical level. Our main
contribution is thus not of a “breakthrough” nature, but innovative in the way we have i)
reordered valuable propositions that were developed in conventional theories of the firm over
the last decades, ii) drawn upon and integrated concepts from a broad range of disciplines
―biology, physics, ecology― with competences more advanced in the field of complexity
science than social sciences, iii) combined the above theoretical-practical propositions and place
them in a methodological sequence to bring the resulting framework to practice, iv) proven the
applicability of the proposed CBVF approach by providing empirical support from real life firms.
In addition to the aforementioned contributions, this thesis harnesses the convergence
of two distinct but interrelated lines of inquiry: theoretical and methodological, which combined
together form a comprehensive approach to tackle complexity in the firm. Such convergence
CONCLUSIONS
224
had shown challenging so far in most literature, due to the lack of a coherent conceptual
framework of complexity focused on behavioral phenomena and sufficient experimental data
for modeling complexity. Meanwhile, our CBVF approach integrates into a single theoretical-
practical framework old and new concepts from otherwise dispersed disciplines, works hands-
on on firms’ behavioral phenomena that is readily tractable, and enables researchers and
practitioners to model and simulate the firm as a complex system, as indicates the empirical
evidence presented in the thesis.
In setting the foundations for the CBVF approach, the author has sought to overcome
some of the epistemological constraints of traditional economic theory ―mainly those based on
the analogy of mechanics, equilibrium and linear behavior― and attempted to explain firms’
behavioral phenomena in a more realistic way than the average conventional theories of the
firm. For example, unlike the conventional approaches of Porter’s Value Chain (Section 4.3.5), or
Penrose’s Resource-based view of the firm (Appendix A), the CBVF introduces a firm-as-a-system
approach that enables researchers and practitioners a greater degree of flexibility in
representing the firm. At the same time, the CBVF incorporates a greater number of elements
from reality into the approach, and it is accompanied by a method that makes the transition to
the practical application of the CBVF smoother and focused on continuously feeding the
assumptions made.
Whilst in the CBVF theory building process, the author has observed how nowadays
factors such as new and more sophisticated computing tools and our increased capacity to
address complexity are dramatically altering our understanding of the behavior of the firm. This
has made us realize that current conventional theories of the firm are at best insufficient and, at
worst, obsolete.
CONCLUSIONS
225
As a result, this author submits that if the heterogeneous body of conventional theories
of the firm is to make further progress in the future, they should essentially get rid of some of its
most conservative assumptions and start to internalize the type of assumptions and methods
covered by a complexity-based approach, such as those presented by the CBVF in this thesis.
What this fundamentally entails for the theory of the firm is, paraphrasing Kauffman’s assertion,
that poised coherence, precarious, subject to avalanches of change in the world of the firm is
inevitable (Kauffman 1990).
To support this undergoing complexity-led change of paradigm, this thesis has
presented and proved the applicability of a tentative complexity-based theory that addresses in
a more realistic way key firm’s behavioral phenomena. The following are some of the main
theoretical conclusions obtained:
1. The firm is an open complex system embedded in an environment from which it
receives and to which it transfers value. No firm exists in isolation from its
environment, but only connected to the outside world. From a complexity point of
view, understanding this principle is a prerequisite to achieve a truly effective and
realistic understanding of the structural and behavioral nature of the firm.
2. Although the open complex system approach challenges some conceptual
conservatism grounded in well-established theories of the firm (Oberheim,
Hoyningen-Huene 2013), it is not totally incompatible with the findings and insights
that we can gather from them. Actually, what the open complex system approach
does is weighing the same content and context differently, in such a way that the
CBVF cannot be divorced from the experimental process by which it is generated
(Glaser, Strauss 2009).
CONCLUSIONS
226
3. Value is considered the basic building block of the firm and, as such, it provides the
brick-and-mortar that shapes value repositories as the firm’s elementary functional
components. Value repositories dynamically interact with one another as nodes
within the firm’s network context, continuously creating and exchanging value with
other value repositories from inside and outside the boundaries of the firm. Not less
importantly, the introduction of value repositories as the fundamental component
within the CBVF approach may provide one workable solution to the problem of
modularity and the problem of designing, coordinating, and managing complex
systems (Ethiraj, Levinthal 2004).
4. The number of value repositories continuously interacting within the firm’s network
context is large, provided that our goal is to attain a thorough understanding of the
behavior of the firm. Empirical complexity thus becomes something inherent to the
nature of the firm, thus making it impossible to assess all the behavioral properties
of the firm at once.
5. The firm operates in a hierarchical (network) context. The emphasis on the linking of
value repositories as a primary task of the firm suggests that the firm should be
conceived as a exchanging function rather than a resource-centered or production
function under the CBVF. Such a concept of the firm should lead to a shift in focus,
away from the control of resources or production towards the integration of
knowledge and the creation of value.
6. Deciphering the firm’s network context and the relationships and interactions
between components is a process that, though intricate, can be satisfactorily
accomplished by using expert-based knowledge and soft computing modeling and
CONCLUSIONS
227
simulation techniques. This basically involves carrying out an iterative process
consisting of measuring-modeling-simulation tasks, which helps discerning between
weak and strong interactions among value repositories.
7. The heterogeneous body of theories of the firm is experiencing increasing
integration of mixed theoretical and methodological paradigms coming from
disciplines other than economics. Consequently, finding a balance between
economical and non-economical explanations of structural complexity, relationships
and interactions among components is the new normal.
8. The integration of qualitative and quantitative analytical tools into hybrid
techniques is likely to be a much more productive course of action for the
advancement of the theory of the firm, rather than continue to sustain the
dominance of hard quantitative methods. Moreover, researchers should be fully
aware that the use of highly sophisticated quantitative methods alone does not
always mean that they are applicable to all research questions (Hoskisson, Hitt et al.
1999).
9. The CBVF is fundamentally a reality-prone approach to understanding the structural
and behavioral dimension of the firm. As such, it produces knowledge about
unobservables and considers legitimate to derive rules from them that can guide
managerial action. It is thus worth noting that this attitude inherent of the CBVF
contrasts with logical positivists who rejects this position (Godfrey, Hill 1995).
10. By further exploring and proving new practical applications of the CBVF approach, it
is expected that researchers and practitioners shall be able to provide better
CONCLUSIONS
228
guidance to real-life firms, without them having to resort to cumbersome and over-
simplifying theoretical frameworks.
Considering the theory building nature of this thesis, the author deliberately has not
engaged in any particular quantitative analysis other than that contained in Chapter 5. The
induction of relationships between the key propositions of the CBVF theory was instead
centered on the understanding of the content and context of the constituents characterizing
complexity in the firm. Thus, it was sufficient that a proposition had an importance rating above
medium to be subject of our analytical interest (Binder, Edwards 2010). The canon for judging
the usefulness of the CBVF theory should therefore be focused on how it has been generated, as
well as its logical consistency, clarity, scope, integration and fit and ability to work (Glaser,
Strauss 2009).
In addition to the theoretical conclusions discussed above, the author considers that the
methodological framework proposed in this thesis is a first attempt aimed at developing a
comprehensive framework for researchers and practitioners to effectively bring the CBVF into
practice. As such, the CBVF methodology requires continuous feedback from practical
outcomes, which in turn stresses the importance of continuously advancing in the development
of a robust and flexible methodology designed to have some form of impact on managerial
actions.
Particularly noteworthy is the flexibility and usability that the CBVF methodology has
demonstrated. Its five consecutive stages, each of which builds on the knowledge gathered by
the preceding stage, confers researchers and practitioners the ability to readily become aware
of complexity and prevent them from falling overwhelmed by full-blown complexity of the firm.
Furthermore, the empirical evidence provided in the thesis stresses the importance of
CONCLUSIONS
229
replicability of the CBVF methodology, thus making it a reliable method for firms of almost any
size operating in any industry, and easily trainable.
Notwithstanding the robustness and soundness shown by the CBVF approach, the
author is well aware of long-standing difficulties to bring complexity into practice, which taken
together, have made the adoption of a complexity-based approach by theorists of the firm fall
behind other developments. The ever-lasting challenge to set a theoretical-practical conception
of complexity being one of the more obvious examples.
In this bewildering situation, the author acknowledges that we should not expect to see
the CBVF integrated into the mainstream economic theory of the firm in the near term.
Nonetheless, this thesis shall contribute to develop new theoretical and practical knowledge on
the applicability of complexity in the firm, disentangle complexity from ambiguity, and show
how promising the opportunities offered by such an approach are in order to increase our
understanding of the firm.
THIS PAGE INTENTIONALLY LEFT BLANK.
FURTHER RESEARCH
231
FURTHER RESEARCH
Although this author is convinced that our CBVF forms part of a new paradigm for
understanding and anticipating the behavior of the firm, he is well aware that such an approach
is too far from being accepted as mainstream. In fact, the entire CBVF is at the early stages of
investigation, with its building process encompassing such an extensive theoretical and practical
background that it deserves systematic further research and elaboration to make it progress.
Due to the scale that a detailed CBVF research programme would have, in this chapter
the author summarizes what he considers are the key areas whose deeper study should
continue to shed light on our understanding of the firm as a complex system, and significantly
enhance the applicability of the CBVF approach in the future.
One of the first areas requiring special attention refers to the constituents of
complexity of the firm. Even though in this thesis the author proposes what he thinks are the
most relevant properties of complexity in the firm, a much deeper investigation would be
needed to verify the linkages and interconnectedness that exist among them, as well as to
uncover new constituents not even considered in this thesis. In the course of this investigation,
the identification of similarities and dissimilarities between the conceptualization of complexity
in the firm and in other social and natural complex systems should contribute to expand our
knowledge and fine-tune the scope of the CBVF.
FURTHER RESEARCH
232
As we have seen, in this thesis we have mainly focused on value repositories as the firm
chief structural components. Further research should next turn our attention to the particular
evolution of value repositories and its inner dynamics. This will necessarily involve building a
larger empirical basis for CBVF analysis, which might help us answer questions such as, How do
firms decide on the number of value repositories, and how do they make decision upon value
repositories? or, How do firm’s decisions affect the interactions among value repositories and
the marketplace itself? Additionally, a base of comparative analysis between the CBVF and other
conventional theories of the firm and quantitative firm modeling methods should allow
researchers and practitioners better set out the advantages of a complexity-based approach,
and foster cross-pollination of ideas among the different firm approaches.
We have examined the implications of value repositories on behavioral phenomena,
however, value repositories ―as the idea of modularity (Ethiraj, Levinthal 2004)― have many
other important implications for management, strategy, organizational coordination, and
financials. Thus, understanding the full implications of value repositories design and knowing the
trade-offs is an area deserving greater focus.
Further research should also look at the role played by key external agents interacting
within the firm network context ―e.g. partners, providers, suppliers, competitors, etc. Certainly
more investigation is needed on the manner and degree to which these agents affect the
complexity of the firm. More specifically, the study of the role of customers ―as the most
prominent source for value creation and exchange― and their relationship with the firm, is an
area of particular interest for the future development and improvement of the CBVF approach.
Another area requiring further research should target on the way key managerial
functions impact value creation. Gathering knowledge on how different functions are involved in
FURTHER RESEARCH
233
the creation and exchange of value is, in fact, central to our understanding of complexity in the
firm. Additionally, it deserves further investigation the manner in which the organizational
processes of the firm affect value creation and, in particular, how the firm’s innovation
processes impact the value repositories.
As we have seen in previous chapters, the methodological framework proposed in this
thesis is a first attempt aimed at developing a comprehensive method for researchers and
practitioners to effectively bring the CBVF into practice. Given its importance, the CBVF
methodology is a key area that deserves further investigation from a multidisciplinary
standpoint, so that we can continue tackling complexity with confidence.
In developing a better methodological framework, much inspiration might come from
disciplines such as physics and biology, both with a proven tradition of dealing with challenging
complexity-related questions. After all, without a creative and in-depth exploration of the
activities, tools and techniques to approach complexity, it is likely that the CBVF would fall short
to explain the firm’s structural components and its dynamic networked relationships, from
where we can later conduct robust modeling, simulating and optimizing processes.
The above would inevitably lead to the need to further investigate and develop CBVF-
specific hybrid tools, which in turn should enable researchers and practitioners to
conceptualize/ characterize complexity of the firm in an increasingly idiosyncratic manner. This
would probably make us reflect on the role of mathematics and quantitative methods in the
study of complexity, as well as to further explore into soft computing techniques to improve our
empirical knowledge of the firm. The path followed by natural sciences decades ago —i.e. when
they had to develop their own analytical tools to cope with complexity— shows us a course of
action that might stimulate the next rounds of CBVF investigations.
FURTHER RESEARCH
234
Particularly relevant in the investigation of the CBVF methodology is to dig deeper into
how the modeling and simulation processes may contribute to advance our understanding of
the complex behavioral dynamics of the firm. For example, the FCM-based model formulated in
this thesis for the airline industry (Chapter 5) has been proved useful for increasing our
understanding of the firm’s behavior, and it seems plausible that it might provide decision-
making capabilities in the near term.
Specifically, FCM models of the firm seem well suited to study feedbacks and test
unintended side effects of diverse firm policy interventions. As such, FCMs might continue to be
a good reference tool to help researchers and practitioners cross the chasm between firm’s
reality, qualitative experts’ knowledge, and quantitative models. Undergoing advances in the
theory and practice of FCM modeling should therefore be brought to CBVF experimentation
―i.e. new algorithms, model extensions. This should call for increased collaboration between
complexity practitioners and FCM computing research teams.
Furthermore, FCM modeling, together with other advanced modeling techniques ―e.g.
neural networks, agent-based modeling― should eventually lead to new hybrid methods that
are used for firm’s scenario generation and simulation. The resulting outcomes might later be
used as a standalone output, or as an input for the generation of value creation roadmaps ― i.e.
to see how different value creation models play out against the future states of the business
environment, or to assess the impact of different policies on performance indicators such as
Operating margin.
Other applications of CBVF modeling might involve creating the basis for further
discussions between C-level executives and decision makers, contrasting simulations against
stakeholders opinions, and/or comparing simulation outcomes with already existing scenarios.
FURTHER RESEARCH
235
Of particular interest would also be to apply the CBVF approach to the domains of a specific
firm, and explore the consequences derived from the absolute measures of firm’s components
on the analysis of complexity.
Although the strong managerial implications and opportunities of the CBVF have been
made evident in this thesis, this does not mean that researchers and practitioners should be
poised to abandon current conventional theories of the firm. Instead, what this author contends
is the idea that, no matter what theory or mindset they choose, they should keep a foot in a
complexity-related view of the firm, such as our CBVF.
A complexity engagement would open new perspectives for structural, strategical and
performance improvements for the firm as a whole and/or its particular value repositories, and
even serve as engine of growth. Ultimately, we must note that our CBVF brings to the fore a life-
long view of the firm in sharp contrast to the hectic short-termism perspective so characteristic
of modern firms; a view that may be used by managers to retain some control over the firm’s
own differential advantage— and thus over its own fate.
THIS PAGE INTENTIONALLY LEFT BLANK.
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
237
APPENDIX A.
REVIEW OF THEORIES OF THE FIRM
A.1 Introduction
This Appendix provides a general overview of the key assumptions made by a selected
group of theories of the firm ― those whose impact this author considers has been more
profound on our current understanding of the firm― and the extent to which they address real
firms’ behavioral problems by tackling complexity. Moreover, the present analysis covers the
key strengths and weaknesses of a heterogeneous body of theories from the last 80 years,
drawing particular attention to the way theoreticians of the firm have approached real
behavioral problems, addressed complexity, and proposed practical guidance.
Note that this is necessarily a short theoretical review not intended to be exhaustive.
The field of the theory of the firm is so vast that a comprehensive assessment would largely
exceed the scope of this thesis. Notwithstanding the foregoing, this author is aware that some
valuable approaches to the firm have not been addressed in this review, the reason being is
because they are either somehow subsumed in the already reviewed theories, or because their
influence over subsequent developments have been not as high as others.
The theories of the firm herein reviewed are the following:
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
238
Neoclassical theory of the firm.
Transaction costs theory.
Agency theory.
Resource-based view of the firm.
Knowledge-based view of the firm.
Stakeholder theory.
Organizational theories of the firm.
Developments in strategy.
A.2 Neoclassical Theory of the Firm
Profit maximization, agent rationality, equilibrium, economies of scale, perfect
competition, and linearity of economic interactions are the key assumptions broadly underlying
most of the neoclassical theory of the firm (Marshall 1919, Pigou 1932, Robinson 1934,
Chamberlin 1949, Marshall, Guillebaud 1961).
In a broad sense, neoclassical scholars think that real business life consisted of the firm
growing until it achieved technical and organizational economies of large-scale production, at
which time, the managerial abilities of the owners of the firm and their desire for ever greater
wealth would wane. The firm would then become increasingly inefficient and, in the end, would
die and make way for new firms led by yet another generation of new managers (Marshall,
Guillebaud 1961, Moss 1984).
The firm in the neoclassical economic theory “is little more than an entrepreneur who is
attached a cost curve or a production function (…) the single participant explicit treated as a
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
239
rational individual” (Mahoney 2005). Other relevant participants of the firm’s ecosystem —such
as the customers, employees, suppliers, partners, etc.— are considered mere “conditions” to
which the entrepreneur adjusts in finding an economic solution that is optimal to him/her .
The neoclassical theory of the firm provides an ideal rational individual who chooses
among fixed and known alternatives, to each of which the known consequences are attached as
if knowledge was fully and perfectly accessible. In fact, the neoclassical theory of the firm does
not deal with the behavior of the firm stricto-sensu, nor does it deal with the firm as an
organization (Dietrich, Krafft 2012). Its chief mission is none other than to “understand how the
price system coordinates the use of resources, not the inner workings of real firms” (Demsetz
1988). Behind these strongly accepted postulates and its brilliant logic, the neoclassical theory
of the firm seems to hide disguised its historical inability to explain some of the central conflicts
and dynamics with which the firm has become increasingly concerned.
As a result, the neoclassical theory of the firm works in a rather mechanistic fashion as
the assumptions of perfect competition and rational agents come into play, with a low level of
contact with empirical data. Moreover, since profit maximization and internal efficiency are
assumed, “there is little room in the neoclassical theory for (…) real business firms" (Simon
1982).
Similarly, neoclassical theory does not “describe the processes that humans use for
making decisions in complex business situations” (Mahoney 2005). The "rational" and "efficient"
behavior with respect to the individual choice is subsequently reduced as a problem of finding
the maximum of some function that is taken (Simon 1982). From a methodological perspective,
this means that the regression paradigm becomes the prevalent analytical tool for most
hypothesis testing in neoclassical thought.
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
240
The central argument of maximizing behavior is no longer supported today, as
researchers have demonstrated that individual perception and cognition intervene between the
decision-maker and the environment, proving neoclassical economics flaws. Instead, a number
of models of satisficing behavior are taking the lead, where maximization is replaced by targets
and satisficing goals, and mechanisms of learning and adaptation.
Decision-makers today are no longer those wise entrepreneurs able to make the best
possible decisions, and we are well aware that our mental models encompass only a fraction of
all the relevant characteristics of a particular firm and of all the information available.
Furthermore, researchers have made available a large mass of descriptive data that
show how most of the key assumptions of the neoclassical theory of the firm do not hold up in
the face of facts. Consequently, the regression paradigm is not the best for testing hypotheses
for all data that is non-experimental and laden with non-recursive relationships.
In contrast with the neoclassical approach to the firm, current firm research is moving
beyond cross-sectional, multiple regression approaches to methods more attuned to the specific
problems and issues likely to influence the behavior of the firm (Hitt, Keats et al. 1998). For
example, to deal with situations too complex for the application of known optimization methods
one of the most widely used techniques is now simulation. In simulation, the trial and error is
supplied by the human investigators rather than by the technique of analysis itself. Other
methods that progressively incorporates complexity as part of the analysis include network
analysis, panel data analysis, logistic regression, structural equation modeling (Hitt, M A et al
1998), as well as the very latest techniques of soft computing.
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
241
In summary, the neoclassical economic theory might be criticized by its conspicuous lack
of empirical testing of assumptions, theories and predictions (Simon 1962a). Neoclassicals have
not only been careless with the real-life complexity of the firm, but have rather preferred to
keep the focus on the efficient allocation of resources based on the marginal analysis and
maximization (Williamson 1996). Not surprisingly, the picture of the firm that comes out of new
research is that of a searching, information processing, satisficing, allocating “mechanism”,
focused on the need to increase its empirical demonstrability (Simon 1962b).
A.3 Transaction Costs Theory
The concept of transaction costs helps explain why the firm exists, and what activities
the firm may undertake or refuse. Originally conceived by Nobel laureate Prof. Ronald H. Coase
in his seminal work “The Nature of the Firm” (Coase 1937), the transaction costs theory provides
an exemplary explanation on how firms would emerge to organize what would otherwise be
market transactions, particularly whenever their costs are less than the costs of carrying out
transactions through the market.
The “complexity” problem in the transaction costs theory is mostly epitomized in the
market mechanism, considered an institution that exist to facilitate exchange; namely to reduce
the costs of carrying out exchange transactions (Coase 1937). Moreover, if we consider
transaction costs somehow analogous to frictions in mechanical systems, the very existence of
the firm might be devised as a mean to respond to a quantifiable balance between price and
costs. As long as using the price mechanism has a cost —e.g. the cost of “organizing” production
through the price system— Coase concludes that it shall be profitable to establish a firm. On the
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
242
contrary, the “diminishing returns to management" makes the bigger companies look less
attractive to carry out activities.
Unlike the neoclassical economic theory, the transaction costs theory characterizes the
firm rather than as a production function, as a governance structure (Williamson 1985), whose
main purpose is economizing on transaction costs. Transactional considerations, nor technology,
nor customers, nor any other constraints from the firm’s outer environment, are decisive in
determining which mode of organization (market vs firm) will score, and in what circumstances
and why (Williamson 1975).
The identification, explanation, and mitigation of contractual hazards are therefore key
assumptions of the transaction costs analysis (Williamson 1996), which helps explain why we
observe so many kinds of organizations. For example, transaction costs theory suggests that
internal production is more likely when the assets are specific and the uncertainties in
contracting are large (Masten 1984).
In transaction costs theory there is an institutional environment —laws, polity— and
institutions of governance —markets, hierarchies— which serve to incorporate a higher degree
of complexity, mostly pertinent to industrial organization. However, this promising intent to
approach the theory to the reality of the firm rapidly falls short. Taking the firm’s institutional
environment as given, the economic agents purportedly align transactions with governance
structures to effect economizing outcomes.
Despite the failed attempt to provide a realistic explanation on the behavior of the firm,
the transaction costs theory substantially increases our understanding of the behavior of the
firm, particularly when compared with the neoclassical economic theory. It also should be noted
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
243
that the transactions costs theory has been particularly fruitful in providing valuable insights in
the analysis of the multidivisional form (M-form) (Chandler 1962, Williamson 1969), and hybrids
form of organization (Hoskisson, Hitt et al. 1999).
A.4 Agency Theory
Whenever one individual depends on the action of another to meet his/her goals, an
agency relationship arises. The individual taking the action is called the agent, whereas the
affected party is the principal (Pratt, Zeckhauser 1985). Challenges in the agency relationship
generally arise whenever the principal cannot perfectly (and costlessly) monitor the agent's
action and information. When this happens, the problems of inducement and enforcement then
come to the fore (Ross 1973).
The key building blocks of the agency theory are the information and economic
incentives (Hölmstrom, Milgrom 1994). This perspective stresses the importance of viewing the
firm as “a system”, specifically as a coherent set of complementary contractual arrangements
which mitigate incentive conflicts (Foss 1998).
The agency theory postulates that there exist information asymmetries between
principals and agents, and that agents typically know more about their tasks than their
principals do. As a consequence, no one can expect a firm to function as if all the information
were costlessly shared, or as if the economic incentives of principals to agents were costlessly
aligned. Regretfully, an agency loss or agency costs invariably emerge (Jensen, Meckling 2000,
Pratt, Zeckhauser 1985).
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
244
Even though agency costs might suggest an overlap with the transaction costs theory,
Williamson (1996) suggests the following three important differences between the agency and
the transaction costs theories:
The units of analysis are different: the individual in the agency theory, and the
transaction in the other
Agency costs focus on ex-ante costs, whereas transaction costs emphasize ex-post
costs
There is a legal centralism assumption of the agency theory, whereas a private
ordering assumption is assumed in the transaction costs theory.
From an operational point of view, the firm’s main economic challenge in structuring an
agency relationship is to minimize the agency costs; or as the agency theory states, to do the
best to achieve what is sometimes called a second-best solution (Mahoney 2005).
Economic incentives are the principal’s means to influence the agents' behavior and to
reap the greatest advantage from the agency relationship, yet reward the agents enough so that
they do not quit. Whether the principal is in a position to design the monitoring and incentive
mechanism, and obtain all the economic benefits from improvements in performance, are two
of the most discussed assumptions that very often are not satisfied (Pratt, Zeckhauser 1985).
As promising as the expectations created by the agency theory were to open up new
opportunities for exploring the firm’s complexity. the truth is that most of the agency theory
debate has revolved around the stale stakeholder-manager dynamics, both in terms of
managerial and philosophical implications. Moreover, the principles of the agency theory have
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
245
been applied to a number of substantive business topics, including innovation, corporate
governance, and diversification (Hoskisson, Hitt et al. 1999).
Agency relationships may well be a pervasive fact of economic life (Arrow 1984),
however, the agency theory fails to provide plausible answers to some of the major firm’s
behavioral questions, as well as to deliver practical guidance to practitioners of the firm, aside
from costs and incentives management.
Nevertheless, a window of opportunity remains open if the agency theory is to be used
to further explain the relationship between the firm and the external agents with which it
interacts. The well-studied one-to-one agency relationship might be leveraged to further explain
the omni-directional relationship space between the firm and its customers, suppliers, partners,
etc. This multi-agent approach —consistent with Arrow’s account such that a single principal
may have many agents (Arrow 1984), and vice versa—could well serve to take a step further in
the complexity of the firm.
Unlike the “canonical” agency theory, where the roles of the principal and the agent
seem perfectly delineated and assigned, under the multi-agent approach the firm might adopt,
at any given time and in relation to any given agent, the role of either a principal or an agent.
From a practical perspective, the roles of the principal and the agent would become two sides of
the same coin, as the adoption of either role would depend on the goal pursued by the firm with
respect to a particular agent, as well as on the cost-benefit rationale involved. This principal-
agent duality of the firm would also entail that:
the firm might play the role of a principal and an agent simultaneously
the firm might be eligible either to pay or receive agency incentives
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
246
the principal is no longer in a position to design the monitoring and incentive
mechanism alone
the principal might not obtain all the economic benefits from the improvements in
the agent’s performance.
The adoption of either a principal or an agent role by the firm in pursuit of greater
efficiency (Levinthal 1988), or for the reduction of uncertainty (Hölmstrom 1979), would
become an acceptable approach for a more complexity-based view of the firm.
A.5 Resource-based View of the Firm
The resource-based view of the firm is a well-grounded stream of research on the
behavior of the firm and business competitiveness. As one of the most influential streams of
firm research of our days, its development is to be found scattered across a large number of
academic references. Yet, we can trace back its foundational ideas to Edith Penrose’s seminal
work “The Theory of the Growth of the Firm” (1959), where she sets the groundbreaking view of
the firm as a bundle of resources.
From Penrose’s pioneering work, passing through the contributions made by
Wernerfelt and Barney in the eighties and nineties (Wernerfelt 1984, Barney 1991), and to our
days, researchers on the research-based view of the firm have mainly focused on two
fundamental guidelines: (1) how differences in firms’ resources realize superior firm
performance, and (2) what and how specific resources give rise to sustainable competitive
advantages ―e.g. learning, culture, entrepreneurship…
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
247
Today, the resource-based view of the firm provides a robust framework for increasing
dialogue between scholars from different disciplines. Moreover, several sub-streams have
emerged from the original view, examples of which include the strategic leadership and the
knowledge-based view of the firm. Such an intense theoretical development has made the
resource-based view one of the prevailing economic perspectives in areas such as strategic
theory and business management.
The resource-based view emphasizes the role of the internal resources of the firm,
particularly the productive services available from management. The general purpose of the
business firm is to organize the use of its "own" resources for the production and sale of goods
and services at a profit, together with other resources acquired from outside the firm (Penrose
1959, Wernerfelt 1984, Barney 1991). From this perspective, the firm becomes a pool of
resources, the utilization of which is organized in an administrative framework. This explains
why both the administrative framework and the administrative decision-making processes are
key elements in this theory.
In the resource-based view, the outer environment of the firm is an "image” in the
entrepreneur's mind of the possibilities and restrictions with which the firm is confronted. In
other words, a given “boundary” that sets the limits of the firm, but which can allegedly be
shaped at will. Consequently, firms have the capacity to alter the conditions of the outer
environment and even to influence consumers’ demand. All that is needed is to make use of an
imaginative effort, develop a good sense of timing and appeal to the instinct for predicting what
will catch on (Penrose 1959). By learning how to manage the specific set of internal resources
and capitalize on the unused productive capacity, the firm may feature the same outer
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
248
environment differently from other firms, thus seizing growth opportunities that others are not
able to create.
The underlying reliance of the theory on the managerial ability to change things go so
far so as to explain the limits to the growth of the firm as a function of the conditions inside the
firm. In fact, while the resource-based view recognizes that a combination of internal attitudes
and external conditions may limit the growth of the firm, the primary limiting factors have to do
with the internal capacities, such as the existing managerial personnel and knowledge. The
firm’s growth and profitability are therefore variables that depend on the managerial capacity
and knowledge that the firm is able to gather —what is known as the “Penrose effect” (Tan,
Mahoney 2005, Goerzen, Beamish 2007).
The managerial ability and the knowledge held by the firm are extended in other
resource-based sub-streams of research by the contribution of intangible assets —i.e. a specific
technology, the accumulated information about the consumers, a brand name, the reputation,
the corporate culture (Itami, Roehl 1991). According to this complementary perspective, the
accumulation of invisible assets becomes critical for future competitiveness, to the extent that
the decisions made regarding intangible assets can affect the firm’s long-term capabilities and
adaptability.
As stated above, Penrose’s resource-based view of the firm heavily draws on the figure
of the entrepreneur. This refers to the individual or group of people providing entrepreneurial
services, no matter what their position or occupational classification are in the firm.
Entrepreneurial services are understood as those contributions to the operations of the firm
related to the introduction and acceptance of new ideas, the acquisition of new managerial
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
249
personnel, changes in the organization of the firm, raising of capital, or making of plans for
expansion (Welter, Smallbone et al. 2012).
Unfortunately Penrose and the later resource-based view community have not dug
deeper as to explain the (critical) process by which the entrepreneur gathers all the knowledge
he/she needs to alter the environment around the firm, nor how the entrepreneur gets insight
into the firm’s market opportunities, nor how to anticipate the consumer acceptance of the
products, nor how to determine the factors that influence most the entrepreneur’s image of the
outer world.
The resource-based view of the firm has continue growing and progressively
incorporated more complexity-related elements of analysis to its scope. Questions such as the
property rights, the contractual nature of the firm, or the notion of decision rights, have
significantly extended the explanatory power of the theory, insofar as they provide the
economic incentives that shape resource allocation (Alchian, Demsetz 1972, Hart, Moore 1990,
Jensen, Meckling 1992, Jensen, Meckling 2000).
For instance, property rights critically affect decision-making regarding resource use,
hence, they are key to understand the economic behavior and performance of the firm. Note
that under this perspective, the firm would become a method of property tenure by which
owners expect to obtain income.
Notwithstanding the foregoing, as scholars become more aware of the complexity of the
firm and our knowledge of the relationship between the firm and its outer environment
increases, there is no longer any certainty as to whether the firm will be able to keep a true
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
250
sustainable competitive advantage alone by holding exclusive property rights over its tangible
and/or intangible resources.
A.6 Knowledge-based Theory of the Firm
What has become known as “the knowledge-based” theory of the firm is the result of
the questions raised in many different fields of inquiry, in the borderline between economics
and business administration, such as strategy research, international business and technology
studies (Foss 1998).
One of the first statements of the knowledge-based approach may be found in
Penrose’s “Limits to the Growth and Size of Firms” (1955). As Penrose explains, firms can be
understood as collections of resources and services derived from these resources, all organized
under an administrative framework. It is through various learning processes, mostly carried out
by the management team, that the firm’s activities are routinized so as to release resources
(Penrose 1955).
After Penrose, a step forward is taken by the introduction of the term “capabilities”
(Richardson 1972), to talk about the necessarily limited range of productive knowledge firms
and individuals possess. According to this perspective, capabilities not only are determinants of
the boundaries of the firm, but they also classify productive activities according to the
capabilities they share.
Other authors later on have focused their attention in exploring the mechanisms
through which firms create knowledge (Grant 1996), conceptualizing the firm as an institution
for integrating knowledge, instead of just applying knowledge. The resulting theory, as Grant
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
251
himself argues, would have implications for the “basis of organizational capabilities, the
principles of organization design (in particular, the analysis of hierarchy and the distribution of
decision-making authority), and the determinants of the horizontal and vertical boundaries of
the firm” (Grant 1996).
The knowledge-based view of the firm is different from other theories of the firm in that
it emphasizes the role of organizational factors in the production of competitive advantage,
primarily focusing on the complex internal organization (Grant 1996, Blomqvist, Kianto 2007). In
contrast with the resource-based view of the firm, more interested in identifying the essential
productive (knowledge) resources and examining how these resources can be acquired and
protected, the knowledge-based approach assumes that knowledge can be managed with tight
procedures, policies, and defined action (Von Krogh, 1998).
Advocates of the knowledge-based approach argue that knowledge is the new
fundamental basis of competition and for the creation of economic value and competitive
advantage. Consequently, knowledge is the most important source of revenue, and it should be
looked at as a distinct form of capital. Given the key relevance of knowledge, firms should adopt
a strategic approach to knowledge (Blomqvist, Kianto 2007).
The knowledge-based view of the firm represents a significant leap ahead in the intent
to provide a more realistic theoretical understanding of the firm when compared with its parent
theories. We could even assert that this approach succeeds in calling our attention to a new
factor such as knowledge, which significantly affects the behavior of the firm. Nonetheless, the
knowledge approach bears the traces of its parent resource-based theory of the firm, from
which it does not depart methodologically.
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
252
A.7 Stakeholder Theory
In the words of Edward Freeman, one of the precursors of the modern stakeholder
theory, a stakeholder is “any group or individual who can affect, or is affected by, the
achievement of a corporation's purpose” (Freeman 1984). According to this definition, the term
“stakeholder” encompasses almost everyone with either a strong or weak connection with the
firm. Given the impractical consequences derived from this definition, Freeman himself later
strove to narrow such definition to “only” include the “employees, customers, suppliers,
stockholders, banks, environmentalists, government and other groups who can help or hurt the
corporation” (Freeman 1984).
The stakeholder thesis is managerial in essence, and was originally conceived to help
corporate managers operate their particular areas of responsibility rather than addressing the
problematic of management theorists and other cross-disciplinary scholars. However, the
original scope of the theory has not prevented dozens of scholars to dig deeper into the
stakeholder theory and make it grow into one of the most prolific areas of management
research of our days.
The focus of the stakeholder theory is articulated around two core questions (Freeman,
Wicks et al. 2004). First, it addresses the question of what the purpose of the firm is. Second, it
inquiries into the responsibility that management have with stakeholders. Both are topics that
delve into the critical process of value creation in the firm and the critical role of the people who
voluntarily come together and cooperate to improve everyone circumstances and create value.
Consequently, for stakeholder theory it is key that managers create relationships with the
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
253
stakeholders so as to build up communities where everyone strives to give their best to deliver
the value the firm promises (Freeman, Wicks et al. 2004).
Stakeholders theorists emphasize that many firms have developed themselves and run
their business in terms consistent with the stakeholder theory, and that by focusing on the
values and relationships with stakeholders they have achieved outstanding performance —e.g.
Google, eBay, J&J…(Collins, Porras 1994). Stakeholder-oriented companies, they claim, are
better positioned to create outstanding customer service than “non-stakeholder companies”, as
they constantly search for ways to cooperate with stakeholders.
For advocates of the stakeholder theory, the future of corporations lays in attending the
legitimate interests of those groups and individuals who can affect, or be affected by, their
activities (Donaldson, Preston 1995). For this purpose, managers need to use good judgment if
they are to create relationships with stakeholders in the right direction (Freeman, Wicks et al.
2004).
Critics of the stakeholder theory have cast doubt on the key assumptions of the theory
and pointed out to its weaknesses. Many authors find it difficult to agree that “companies
perform better the more closely they engage everyone affected by their operations (and that, as
a result, companies) should be run for the benefit of all those who may be deemed to have a
stake in it” (Argenti 1997). These authors find hard to decide who is a stakeholder and who is
not. Moreover, they do not easily accept why the stakeholder theory does not provide a clear
idea of what stakeholders should expect to receive, and they argue that it is unrealistic to expect
managers to make a trade-off between stakeholder interests, unless the theory interferes with
the firm’s performance.
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
254
Instead, critics claim that what managers really do is to ask themselves “what effect
each decision might have on profits” (Argenti 1997), which means that the stakeholders should
be treated according to the effect they have on profits. Furthermore, they argue that companies
need to behave towards everyone in such a manner that they benefit from its activities and
become “collateral beneficiaries”.
A large part of the debate between advocates and critics of the stakeholder theory has
recently focused on whether companies should focus on the stakeholders or the shareholders
instead, as well as on discussing the purpose of the company itself (Sundaram, Inkpen 2004).
Literature shows multiple examples in favour of one or the other side in what it seems to be a
never-ending debate. But there is also common ground between the two sides. For example,
they seem to agree is that companies need to have a clear purpose and cannot survive unless
they deliver true value to their chosen stakeholders (Campbell 1997).
At the end, this enriching debate highlights the need for a more complexity-based view
of the firm, the development of which will be possible as the stakeholder theory evolves into a
more holistic approach to the relationships between the firm and the actors in its ecosystem.
A.8 Organizational Theories of the Firm
The organizational theories of the firm deal with the fact that there are practical limits
to human rationality, and that these limits are not static, but depend upon the organizational
environment in which individuals’ decisions take place (Scott, Davis 2007). Organization and
rationality are therefore two intertwined concepts that provide feedback one to another.
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
255
According to this approach, given the large number of alternatives and the huge amount
of information that every individual must assess in real business life, objective rationality
becomes something hard to sustain. Therefore, the goal of the organization is to design an
environment of decision such that individuals are rational in their decisions —the organization
as a system of cooperative behavior (Simon 1953).
For those researchers operating from the behavioral perspective, the organization is
viewed as a more efficient information processor than the individual (Cyert, March 1963), still
other authors point to the sociological significance of the firm —because of their ubiquity, their
impact on power and status, and as complex set of social processes (Scott, Davis 2007). And the
discussion might continue, including those authors arguing that organizations are a means of
achieving the benefits of collective action in situations in which the price system fails (Arrow
1974), or those arguing that the firm exists because they can achieve co-ordination more
effectively than the market so that investment in managerial hierarchy achieves productivity
gains —Chandler’s concept of the ‘visible hand’ (Chandler 1977, Chandler 1990).
Whatever the perspective is, the term “organization” endorses the fact that virtually all
decisions require the participation of many individuals for their effectiveness. Therefore, for
organizations to work effectively they must be provided with employees and a pattern of human
communication and relationships. Employees are key stakeholders for the outcome of
organizations, and their work is carried out within the terms of an authority relationship
between the employer and the employees. Both the economic relationship created by the
employment contract and the compensation framework, play a central role in organizational
theories (Simon 1982).
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
256
The organizational theories of the firm are closely related to the principles of the
evolutionary theory, as specified in Nelson and Winter’s seminal work “An Evolutionary Theory
of Economic Change” (1982), where a set of ideas are provided as to the way organizations
evolve and the role the environment plays in economic change.
Nelson and Winter argue that much of the firm’s behavior can be understood as a
reflection of routines and strategic orientations. In their evolutionary theory, routines play the
same role that genes play in biological evolutionary theory. They are a persistent feature of the
organism and determine its possible behavior, even better than if the firm carries out a
thorough decision-making process into the future (Nelson, Winter 1982).
According to these authors, most of what is regular and predictable in the firm’s
behavior is subsumed under the term (organizational) "routine". Consequently, the non-regular
and unpredictable behavior is furnished by recognizing that there are stochastic elements both
in the determination of decisions and of the decision outcomes themselves. This conception of
stochasticity in the behavior of the firm holds the ability to alter “inheritance” of the firm’s
characteristics, and provoke the appearance of organizational variations (or mutations).
Under Nelson and Winter’s evolutionary theory, firms are motivated by profitability and
engaged in search for ways to constantly improve their profitability. For this purpose, firms have
specific organizational capabilities and decision rules, at any given time (Nelson, Winter 1982).
Over time, these organizational capabilities and decision rules are modified as a result of both
deliberate problem-solving efforts and stochastic events coming from outside the boundaries of
the firm. As it can be observed, the economic analogue of natural selection operates as the
market determines which firms are profitable and which are unprofitable, and tends to winnow
out the unprofitable firms.
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
257
Last but not least, it is worth highlighting the conceptualization of the organization as an
open system, a perspective from which our complexity-based view of the firm sources many
ideas. First introduced by K.E. Boulding in 1956, the open system approach represents an
important breakaway from the mechanical models too often presented in the theory of
organizations. As Boulding argues:
“(…) in dealing with (…) organizations we are dealing with systems in
the empirical world far beyond our ability to formulate. We should not be
wholly surprised, therefore, if our simpler systems, for all their importance
and validity, occasionally let us down” (Boulding 1956).
After Boulding, the open system approach research agenda has continued to grow,
inspiring dozens of authors more deeply to investigate the open system view of the firm. Of
particular interest for our complexity-based view of the firm is Sanchez and Heene (2004)
perspective of the organization as a goal-seeking open system, which aims to create and
distribute value. According to these authors, the organization has important interactions across
its boundaries between the resources within the organization and those in its environment, to
the extent that the organization takes into account or is influenced by its own influence on the
environment (Sanchez, Heene 2004).
New approaches in the field of the organizational theories of the firm —contingency
theory, learning organization, etc.— go beyond simple determinism to models that include
simultaneity, holism, emergence, real-life knowledge, flexibility, and the need for strategic
choices. These new approaches presume evolving conflicts between external contingencies and
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
258
internal consistency (Maula 2006), which is a significant step further towards a complexity-
based view of the firm.
A.9 Developments in Strategy
The most recent developments in the theory of the firm incorporate work on firm
strategy and management, which reflect a more “applied” character and the strengthen
dialogue between “pure” and “applied” fields.
As such, the interconnectedness between the firm and its outer environment strongly
gains importance, with several authors leading the way in the literature.
Worth mentioning is the work of Kapoor and Lee (2013) on US hospitals, who consider
firms in the context of their business ecosystems and explore how governance choices with
respect to “complementors” and “distributors” shape their competitive behavior. According to
these authors, governance choices play an important role in the firm’s ability to coordinate
changes in interdependent activities so as to create value (Kapoor, Lee 2013). Their work
anticipates the impact on organizational forms of the coordination mechanisms and competitive
behavior, the choice of which must be considered in the context of business ecosystems (Iansiti,
Levien 2004, Kapoor, Lee 2013).
Also relevant, and complementary to the above, is the examination of shakeouts in the
context of business ecosystems and, more precisely, the market turbulence generated by core
firm decisions in ecosystems. As some authors show, firms’ decisions can generate financial
losses and exit for niche market firms. Pierce’s study on the US automotive industry, highlights
the importance of environmental and firm characteristics in shakeouts, and goes as far as to
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
259
develop hypotheses predicting which niche markets will suffer larger losses and be more
susceptible to shakeouts, and how core firm decisions will drive complementor performance
and survival (Pierce 2008).
These worthwhile attempts to incorporate a more deeper understanding of the firm’s
environment into the core of the theory of the firm are complemented by other novel value-
centered approaches, such as Porter et al’s “Creating Shared Value” (2007). As these authors
state:
“A big part of the problem lies with companies themselves, which
remain trapped in an outdated approach to value creation that has emerged
over the past few decades. They continue to view value creation narrowly,
optimizing short-term financial performance in a bubble while missing the
most important customer needs and ignoring the broader influences that
determine their longer-term success” (Porter, Kramer 2011)
Thinking about the possible solutions to help capitalism emerge from the crisis in which
it finds itself today, these authors emphasize the role of the firms to “bringing business and
society back together”. The solution lies in the principle of shared value, which involves creating
economic value —not just profit per se— in a way that also creates value for society by
addressing its needs and challenges:
“Shared value, then, is not about personal values. Nor is it about
“sharing” the value already created by firms—a redistribution approach.
Instead, it is about expanding the total pool of economic and social value”
(Porter, Kramer 2011)
APPENDIX A. REVIEW OF THEORIES OF THE FIRM
260
Although sometimes presented as the next big idea by the strategy management circles,
Porter et al.’s notion of “shared value” is more conceptual than practical and resembles many of
the ideas discussed in the more mature stakeholder theory.
APPENDIX B. DELPHI’S EXPERT BENCHMARKING OF RESPONSES
261
APPENDIX B.
DELPHI’S EXPERT BENCHMARKING OF
RESPONSES
In this Appendix we provide an example of an individual report containing a
benchmarking of the responses from an expert in Round 3 against the aggregate responses from
the Experts’ Panel. A report containing the information contained in this Appendix was
distributed individually to all Panel members in Round 4 of the Delphi process, and used as a
basis to respond the questions set in Round 4 questionnaire (see Appendix D).
In particular, the figures presented in this Appendix include:
A benchmark of an expert’s responses vs Panel members responses asking for the
interconnectedness between constraints and value repositories (Fig.30).
A benchmark of an expert’s responses vs Panel members responses asking for the
interconnectedness among value repositories (Fig.31).
A benchmark of an expert’s responses vs Panel members responses asking for the
interconnectedness between value repositories and Operating Margin (Fig.32).
The data contained in these figures graphically illustrate the differences between
expert’s responses and those of the Panel members in Round 3, the latter represented as the
APPENDIX B. DELPHI’S EXPERT BENCHMARKING OF RESPONSES
262
average values of all experts’ responses. This information offered every expert participating in
Delphi Round 4 the opportunity to assess his/her previous level of agreement or disagreement
with Panel members.
B.1 Interconnectedness between Constraints and Value Repositories
As shown in Fig.30, the most notorious differences ―i.e. those at or above 70%
difference rate― between expert’s responses and those of the Panel members are the
following:
Constraints vs Customer experience, 100% difference rate.
Constraints vs People and talent, 90%.
Constraints vs Brand, 80%.
Constraints vs Capacity management, 70%.
Constraints vs Safety and security, 70%.
Constraints vs Information management, 70%.
If we analyze where the smallest differences are given, namely where there has been a
higher level of coincidence between the expert and the Panel members ―i.e. less than 50%
difference rate― the results obtained are the following:
Constraints vs Network, 40% difference rate.
Constraints vs Alliances, 40%.
Constraints vs Customer centric proposition, 30%.
APPENDIX B. DELPHI’S EXPERT BENCHMARKING OF RESPONSES
263
Figure 30. Benchmarking of Responses: Constraints vs Value Repositories
Source: own elaboration
B.2 Interconnectedness among Value Repositories
Fig.31 shows the high level of coincidence between the expert’s responses and those of
the Panel members, as reflected by the high number of responses whose difference rate is at or
below 20%:
Corporate culture, 13% difference rate.
Distribution strategy, 13%.
Management (leadership), 20%.
Innovation, 20%.
APPENDIX B. DELPHI’S EXPERT BENCHMARKING OF RESPONSES
264
Information management, 20%.
Alliances, 27%.
Customer centric proposition, 27%.
Moreover, when we look at the highest differences in response, the difference rates
remain between 50-40%, with the highest difference in 53%, as shown by the cases below:
Value repositories vs Network, 53% difference rate.
People and talent, 47%.
Capacity management, 47%.
Process and cost optimization, 47%.
Figure 31. Benchmarking of Responses: Value Repositories
Source: own elaboration
APPENDIX B. DELPHI’S EXPERT BENCHMARKING OF RESPONSES
265
B.3 Interconnectedness between Value Repositories and Operating
Margin
The benchmarking of responses on the interconnectedness between Value Repositories
and airlines’ Operating Margin is presented in Fig.32.
Figure 32. Benchmarking of Responses: Value Repositories vs Operating Margin
Source: own elaboration
As we can see, the difference rate between the expert’s responses and those of the
Panel members is 67%. This means that 10 out of 15 questions were answered differently
between the expert and Panel members. The most noticeable difference refers to the
interconnectedness between Alliances and Operating Margin, which was labelled as “non-
existent” by the expert, whereas the Panel members labelled it as “Strong”.
THIS PAGE INTENTIONALLY LEFT BLANK.
APPENDIX C. PANEL MEMBERS
267
APPENDIX C.
PANEL MEMBERS
The Experts’ Panel was the primary participatory group across the thesis field research.
The Panel was created on an on-line basis, and gathered a number of experts from different
network airlines, positions and geographies across the world. All Panel members were
specifically invited for the occasion.
The following table contains the list of the Panel members, including their names,
position and name of company:
Name
Position
Company
Mr. Amin Abdulhadi
Director Flight Operations
Engineering & Administration
Scandinavian Airlines
Mr. Thomas Bartsch
Former SVP Commercial
Intelligence
Qatar Airways
Mrs. Daniela Baytelman
VP Distribution & Ancillary
Revenue
LATAM Airlines
Mr. Dimitris Bountolos
VP Customer Experience
Iberia Líneas Aéreas
Mr. Duncan Bureau
VP Global Sales and Distribution
Air Canada
Mr. Anthony Doyle
Director New Product
Development
Air Canada
Mr. Ferrán García
Strategic Planning & CEO Office
Iberia Líneas Aéreas
Mr. Raúl Gutiérrez
VP & Chief Information Officer,
Club Premier
Aeromexico
APPENDIX C. PANEL MEMBERS
268
Name
Position
Company
Mr. Hans Gydal
Director Sales Excellence
Scandinavian Airlines
Mr. Martin Hoffman
VP Value Stream Owner - Plan to
Execution
Scadinavian Airlines
Mr. Carlos Jovel
VP Revenue Management and
Pricing
LATAM Airlines
Mr. Rahul Kucheria
Head of Loyalty
Qatar Airways
Mr. Rodrigo Llaguno
VP Customer Experience
Avianca
Mr. Juan Felipe Luque
Director of Cargo Operations
Avianca
Mr. Jiří Marek
Executive Director Sales &
Distribution
LOT Polish Airlines
Mr. Juan Alberto Martín
VP Joint Business Agreements
Iberia Líneas Aéreas
Mr. Rafael Andrés Martínez
Head of Distribution & Revenue
Mmgt
Aerolíneas Argentinas
Mr. Albert Muntané
Head of Network and Distribution
Air Europa
Mr. Mark Nasr
MD Corporate Strategy and
Development
United Airlines
Mr. Mauro Oretti
VP Sales & Marketing
SkyTeam
Mr. Ole Orvér
SVP Network Management
Qatar Airways
Mr. Mariano Salinas
Strategy & Business Development
Director
Avianca
Mr. Federico Soto
Former Head of Strategic
Management Office
Iberia Líneas Aéreas
Mr. Maarten van der Lei
VP Pricing & Revenue Mmgt
Europe KLM
AirFrance/KLM
Mr. Warner van der Veer-Jehee
VP Safety & Quality at KLM E&M
Division
AirFrance/KLM
Mr. Helmut Woelfel
VP Commercial
Lufthansa
Mr. Maarten van der Lei
VP Pricing & Revenue Mmgt
Europe KLM
AirFrance/KLM
Mr. Warner van der Veer-Jehee
VP Safety & Quality at KLM E&M
Division
AirFrance/KLM
APPENDIX C. PANEL MEMBERS
269
Name
Position
Company
Mr. Helmut Woelfel
VP Commercial
Lufthansa
CEO: Chief Executive Officer; SVP: Senior Vice-President; VP: Vice-President
Table 19. List of Panel members
Source: own elaboration
Some of the Panel members chose not to make their personal profiles public. The
following table contains less detailed information about these members.
Position
Type of company
Geography
VP Network Operations
Network Airline
Middle East
VP Aircraft Maintenance
Network Airline
Europe
Director Customer Experience
Alliance-Brand
Worldwide
VP Customer Experience and
Technology
Alliance-Brand
Worldwide
VP: Vice-President
Table 20. List of Non-disclosed Panel members
Source: own elaboration
THIS PAGE INTENTIONALLY LEFT BLANK.
APPENDIX D. DELPHI QUESTIONNAIRES
271
APPENDIX D.
DELPHI QUESTIONNAIRES
D.1 Delphi Round 1 Questionnaire
APPENDIX D. DELPHI QUESTIONNAIRES
272
APPENDIX D. DELPHI QUESTIONNAIRES
273
D.2 Delphi Round 2 Questionnaire
APPENDIX D. DELPHI QUESTIONNAIRES
274
APPENDIX D. DELPHI QUESTIONNAIRES
275
D.3 Delphi Round 3 Questionnaire
APPENDIX D. DELPHI QUESTIONNAIRES
276
(pages 5-14 are not displayed)
(pages 17-35 are not displayed)
APPENDIX D. DELPHI QUESTIONNAIRES
277
D.4 Delphi Round 4 Questionnaire
APPENDIX D. DELPHI QUESTIONNAIRES
278
(pages 5-15 are not displayed)
(pages 18-40 are not displayed)
APPENDIX E. FIELD RESEARCH DATA
279
APPENDIX E.
FIELD RESEARCH DATA
E.1 Participants Profile (participants.xlsx)
Expert
Area
Geography
Position
1
Engineering
Europe
Director
2
Commercial
Middle East
SVP
3
Sales & Distribution
South America
VP
4
Commercial
Europe
VP
5
Sales & Distribution
North America
VP
6
Network & Revenue
South America
Director
7
Commercial
North America
Director
8
Corporate
Europe
VP
9
Corporate
Europe
Manager
10
Sales & Distribution
Europe
Director
11
Commercial
Europe
VP
12
Network & Revenue
South America
VP
13
Commercial
Middle East
Manager
14
Commercial
South America
VP
15
Network & Revenue
Middle East
VP
16
Commercial
North America
Director
APPENDIX E. FIELD RESEARCH DATA
280
Expert
Area
Geography
Position
17
Sales & Distribution
Europe
Director
18
Commercial
Europe
VP
19
Network & Revenue
South America
VP
20
Network & Revenue
Europe
Director
21
Corporate
North America
Director
22
Sales & Distribution
Europe
VP
23
Network & Revenue
Middle East
SVP
24
Corporate
Europe
Manager
25
Network & Revenue
Europe
VP
26
Engineering
Europe
VP
27
Engineering
Europe
VP
28
Commercial
Europe
VP
E.2 Participation Data (participants.xlsx)
Expert
Type
Geography
Round
1
Complete
Europe
R1
2
Complete
Middle East
R1
2
Complete
Middle East
R2
2
Complete
Middle East
R3
2
Complete
Middle East
R4
3
Complete
South America
R1
3
Complete
South America
R2
3
Complete
South America
R3
4
Complete
Europe
R1
5
Complete
North America
R2
APPENDIX E. FIELD RESEARCH DATA
281
Expert
Type
Geography
Round
6
Complete
South America
R3
7
Complete
North America
R1
7
Complete
North America
R2
7
Complete
North America
R3
7
Complete
North America
R4
8
Incomplete
Europe
R1
8
Complete
Europe
R2
8
Complete
Europe
R3
8
Incomplete
Europe
R4
9
Complete
Europe
R1
9
Complete
Europe
R2
9
Complete
Europe
R3
9
Incomplete
Europe
R4
10
Complete
Europe
R2
11
Incomplete
Europe
R1
12
Complete
South America
R1
13
Incomplete
Middle East
R1
14
Complete
South America
R1
14
Complete
South America
R2
14
Complete
South America
R3
15
Complete
Middle East
R1
16
Complete
North America
R2
17
Complete
Europe
R1
18
Complete
Europe
R1
18
Complete
Europe
R2
18
Complete
Europe
R3
18
Complete
Europe
R4
APPENDIX E. FIELD RESEARCH DATA
282
Expert
Type
Geography
Round
19
Complete
South America
R1
19
Complete
South America
R2
19
Complete
South America
R3
19
Complete
South America
R4
20
Complete
Europe
R1
20
Complete
Europe
R2
20
Complete
Europe
R3
20
Complete
Europe
R4
21
Complete
North America
R2
22
Complete
Europe
R1
22
Complete
Europe
R2
22
Complete
Europe
R3
22
Complete
Europe
R4
23
Complete
Middle East
R2
24
Complete
Europe
R1
24
Complete
Europe
R2
24
Complete
Europe
R3
24
Complete
Europe
R4
25
Incomplete
Europe
R1
26
Incomplete
Europe
R1
27
Incomplete
Europe
R1
27
Complete
Europe
R2
27
Complete
Europe
R3
27
Incomplete
Europe
R4
28
Incomplete
Europe
R1
28
Complete
Europe
R2
100
Not valid
Unknown
R1
APPENDIX E. FIELD RESEARCH DATA
283
Expert
Type
Geography
Round
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R1
100
Not valid
Unknown
R2
100
Not valid
Unknown
R2
100
Not valid
Unknown
R2
100
Not valid
Unknown
R2
100
Not valid
Unknown
R2
100
Not valid
Unknown
R2
100
Not valid
Unknown
R2
100
Not valid
Unknown
R3
100
Not valid
Unknown
R3
100
Not valid
Unknown
R3
100
Not valid
Unknown
R3
100
Not valid
Unknown
R3
100
Not valid
Unknown
R3
100
Not valid
Unknown
R4
APPENDIX E. FIELD RESEARCH DATA
284
E.3 Delphi Round 1 Constraints (round_1_R.xlsx)
Constraint
Frequency
Labor costs
18
Competition from other airlines
16
Government regulation
15
Fuel cost
14
Airport fees
8
GDS fees
6
Leisure travel demand
5
Business travel demand
4
ATC fees
3
IT systems costs and complexity
2
Handling fees
2
IT interoperability problems
1
Aircraft cost & efficiency
1
Slot availability
1
Economic cycle
1
Competition from other means of transport
1
Power of unions/labour force
1
Government subsidies
1
Capital intensity
1
High demand volatility and price elasticity
1
Inflexible, time-lagged supply
1
Short-term variable costs
1
Price transparency
1
Commoditizied product offering
1
APPENDIX E. FIELD RESEARCH DATA
285
Constraint
Frequency
Duopolio of aircraft manufacturers
1
ATC co-ordination
1
Airport infrastructure
1
New Distribution Capability (NDC) Integration
1
Environmental awareness
1
Excess capacity
1
Currency exchange rates
1
Distortion of global competition due to the
intervention of regional/local bodies, or the lack of
it
1
Government taxation
1
Health threats
1
Cost of sales (including transaction fees and fraud)
1
Access to talent and ability to retain talent
1
Freedom of movement over borders, visa policies
and processing delays
1
Airport security screening and processing
procedures
1
Product distribution capabilities/costs which
includes but expands beyond the GDS costs
1
E.4 Delphi Round 1 Value Repositories (round_1_R.xlsx)
Value Repository
Frequency
Capacity management
6
Information management
5
Network
5
Customer experience
4
Scheduling
3
APPENDIX E. FIELD RESEARCH DATA
286
Value Repository
Frequency
Sales
3
Procurement
3
Operations management
3
Relationships with stakeholders
3
Corporate culture
3
Products and services
3
People and talent
3
Process and cost optimization
3
Digital channels
2
Alliances
2
Revenue management
2
Distribution strategy
2
Customer centric proposition
2
Frequent flyer programmes
2
Flight and cabin crew rostering
2
Finance
2
Customer data analytics
2
Brand
2
Social ecosystem
2
Ancillary revenue
2
Management/ Leadership
2
IT Management
1
Safety and security
1
Innovation
1
APPENDIX E. FIELD RESEARCH DATA
287
E.5 Delphi Round 2 Constraints (round_2_R.xlsx)
Constraint
Frequency
Adj. Frequency
Government regulation
15
15,00002
Fuel cost
11
11,000018
Competition from other airlines
11
11,00001
Commoditizied product offering
11
11,000009
Power of unions/labour force
10
10,000031
Labor costs
10
10,000028
Slot availability
8
8,000034
Excess capacity
7
7,000017
Capital intensity
7
7,000008
Business travel demand
7
7,000007
Airport infrastructure
7
7,000004
Aircraft cost and efficiency
7
7,000002
High demand volatility and price
elasticity
6
6,000024
GDS fees
6
6,000019
Economic cycle
6
6,000015
Airport fees
6
6,000003
Government taxation
5
5,000022
Government subsidies
5
5,000021
Distortion of global competition
due to the intervention of
regional/local public bodies, or
the lack of it
5
5,000013
Duopoly of aircraft
manufacturers
4
4,000014
Price transparency
3
3,000032
Leisure travel demand
3
3,000029
APPENDIX E. FIELD RESEARCH DATA
288
Constraint
Frequency
Adj. Frequency
IT systems costs and complexity
3
3,000027
IT interoperability problems
3
3,000026
Currency exchange rates
3
3,000012
Competition from other means
of transport
3
3,000011
Inflexible, time-lagged supply
2
2,000025
Handling fees
2
2,000023
ATC co-ordination
2
2,000005
Short-term variables costs
1
1,000033
ATC fees
1
1,000006
New Distribution Capability
(NDC) integration
0
0,00003
Environmental awareness
0
0,000016
E.6 Delphi Round 2 Value Repositories (round_2_R.xlsx)
Value Repository
Frequency
Adj. Frequency
Network
17
17,000019
Revenue management
16
16,000026
People and talent
16
16,000021
Management/ Leadership
15
15,000018
Corporate culture
14
14,000006
Capacity management
14
14,000005
Customer experience
13
13,000008
Innovation
12
12,000016
Brand
12
12,000004
Alliances
12
12,000002
APPENDIX E. FIELD RESEARCH DATA
289
Value Repository
Frequency
Adj. Frequency
Safety and security
11
11,000027
Distribution strategy
11
11,000011
Process and cost optimization
10
10,000022
Information management
10
10,000015
Customer centric proposition
10
10,000009
Frequent Flyer Programmes
9
9,000014
Ancillary revenue
9
9,000003
Scheduling
8
8,000029
Operations management
8
8,00002
Sales
7
7,000028
Products and services
7
7,000024
IT management
6
6,000017
Finance
5
5,000012
Customer data analytics
5
5,000007
Digital channels
4
4,00001
Relationships with stakeholders
3
3,000025
Procurement
3
3,000023
Flight and cabin crew rostering
2
2,000013
Social ecosystem
0
0,00003
E.7 Delphi Rounds 3 and 4 Data
Delphi Rounds 3 and 4 data files are not presented here due to the large size of the files.
Please refer to the following files in the accompanying CD for more detail:
round_3_cons_R.csv
APPENDIX E. FIELD RESEARCH DATA
290
round_3_vr_R.csv
round_3_om_R.csv
round_4_conssi_R.csv
round_4_consst_R.csv
round_4_vrsi_R.csv
round_4_vrst_R.csv
round_4_omsi_R.csv
round_4_omst_R.csv
E.8 Edges Values and Attributes (edgespanel.csv)
From
To
Weight
Regulation
Network
-0.65
Fuel
Network
-0.6
Competition
Network
-0.9
Commoditization
Network
0.35
Unions
Network
-0.4
Labor
Network
-0.45
Slots
Network
0.7
ExCapacity
Network
-0.75
Capital
Network
-0.6
Biz-demand
Network
0.83
Regulation
Revenue
0.35
Fuel
Revenue
0.55
Competition
Revenue
-0.98
Commoditization
Revenue
0.5
APPENDIX E. FIELD RESEARCH DATA
291
From
To
Weight
Unions
Revenue
0.33
Labor
Revenue
0.48
Slots
Revenue
0.3
ExCapacity
Revenue
-0.95
Capital
Revenue
0.13
Biz-demand
Revenue
0.95
Regulation
People
0.2
Fuel
People
0.1
Competition
People
0.68
Commoditization
People
0.23
Unions
People
-0.58
Labor
People
0.75
Slots
People
0.18
ExCapacity
People
0.18
Capital
People
0.2
Biz-demand
People
0.15
Regulation
Management
0.45
Fuel
Management
0.3
Competition
Management
0.78
Commoditization
Management
0.53
Unions
Management
-0.6
Labor
Management
0.68
Slots
Management
0.35
ExCapacity
Management
0.4
Capital
Management
0.45
Biz-demand
Management
0.23
Regulation
Culture
0.23
APPENDIX E. FIELD RESEARCH DATA
292
From
To
Weight
Fuel
Culture
0.2
Competition
Culture
0.75
Commoditization
Culture
0.43
Unions
Culture
-0.78
Labor
Culture
0.65
Slots
Culture
0.2
ExCapacity
Culture
0.28
Capital
Culture
0.28
Biz-demand
Culture
0.35
Regulation
Capacity
0.65
Fuel
Capacity
0.83
Competition
Capacity
0.93
Commoditization
Capacity
0.63
Unions
Capacity
-0.3
Labor
Capacity
-0.38
Slots
Capacity
0.7
ExCapacity
Capacity
-0.9
Capital
Capacity
0.63
Biz-demand
Capacity
0.9
Regulation
Experience
-0.2
Fuel
Experience
-0.25
Competition
Experience
0.7
Commoditization
Experience
-0.78
Unions
Experience
0.3
Labor
Experience
-0.33
Slots
Experience
0.3
ExCapacity
Experience
0.33
APPENDIX E. FIELD RESEARCH DATA
293
From
To
Weight
Capital
Experience
0.23
Biz-demand
Experience
0.65
Regulation
Innovation
-0.63
Fuel
Innovation
0.65
Competition
Innovation
0.88
Commoditization
Innovation
0.68
Unions
Innovation
-0.4
Labor
Innovation
-0.4
Slots
Innovation
0.33
ExCapacity
Innovation
0.45
Capital
Innovation
0.65
Biz-demand
Innovation
0.73
Regulation
Brand
0.08
Fuel
Brand
0.18
Competition
Brand
0.88
Commoditization
Brand
0.83
Unions
Brand
0.3
Labor
Brand
0.2
Slots
Brand
0.1
ExCapacity
Brand
0.25
Capital
Brand
0.2
Biz-demand
Brand
0.83
Regulation
Alliances
-0.78
Fuel
Alliances
0.15
Competition
Alliances
0.83
Commoditization
Alliances
0.65
Unions
Alliances
0.3
APPENDIX E. FIELD RESEARCH DATA
294
From
To
Weight
Labor
Alliances
0.38
Slots
Alliances
0.53
ExCapacity
Alliances
0.6
Capital
Alliances
0.15
Biz-demand
Alliances
0.75
Regulation
Safety
0.95
Fuel
Safety
0.05
Competition
Safety
0.13
Commoditization
Safety
0.15
Unions
Safety
0.4
Labor
Safety
0.15
Slots
Safety
0.08
ExCapacity
Safety
0.15
Capital
Safety
0.23
Biz-demand
Safety
0.08
Regulation
Distribution
-0.58
Fuel
Distribution
0.23
Competition
Distribution
0.88
Commoditization
Distribution
0.83
Unions
Distribution
0.15
Labor
Distribution
0.1
Slots
Distribution
0.33
ExCapacity
Distribution
0.5
Capital
Distribution
0.3
Biz-demand
Distribution
0.78
Regulation
Optimization
0.63
Fuel
Optimization
0.75
APPENDIX E. FIELD RESEARCH DATA
295
From
To
Weight
Competition
Optimization
0.7
Commoditization
Optimization
0.63
Unions
Optimization
-0.85
Labor
Optimization
-0.8
Slots
Optimization
0.28
ExCapacity
Optimization
0.65
Capital
Optimization
0.58
Biz-demand
Optimization
0.35
Regulation
Information
0.33
Fuel
Information
0.25
Competition
Information
0.73
Commoditization
Information
0.53
Unions
Information
0.3
Labor
Information
0.18
Slots
Information
0.08
ExCapacity
Information
0.2
Capital
Information
0.23
Biz-demand
Information
0.45
Regulation
Customer-centric
-0.48
Fuel
Customer-centric
0.35
Competition
Customer-centric
0.68
Commoditization
Customer-centric
0.7
Unions
Customer-centric
0.2
Labor
Customer-centric
0.13
Slots
Customer-centric
0.4
ExCapacity
Customer-centric
0.25
Capital
Customer-centric
0.15
APPENDIX E. FIELD RESEARCH DATA
296
From
To
Weight
Biz-demand
Customer-centric
0.83
Network
Network
0.7
Revenue
Network
0.93
People
Network
0.38
Management
Network
0.73
Culture
Network
0.58
Capacity
Network
0.93
Experience
Network
0.53
Innovation
Network
0.4
Brand
Network
0.33
Alliances
Network
0.93
Safety
Network
0.18
Distribution
Network
0.73
Optimization
Network
0.6
Information
Network
0.6
Customer-centric
Network
0.73
Network
Revenue
0.93
Revenue
Revenue
0.55
People
Revenue
0.63
Management
Revenue
0.73
Culture
Revenue
0.58
Capacity
Revenue
1
Experience
Revenue
0.48
Innovation
Revenue
0.58
Brand
Revenue
0.5
Alliances
Revenue
0.85
Safety
Revenue
0.23
APPENDIX E. FIELD RESEARCH DATA
297
From
To
Weight
Distribution
Revenue
0.8
Optimization
Revenue
0.65
Information
Revenue
0.88
Customer-centric
Revenue
0.58
Network
People
0.3
Revenue
People
0.45
People
People
0.7
Management
People
1
Culture
People
1
Capacity
People
0.25
Experience
People
0.85
Innovation
People
0.75
Brand
People
0.68
Alliances
People
0.28
Safety
People
0.3
Distribution
People
0.35
Optimization
People
0.78
Information
People
0.55
Customer-centric
People
0.85
Network
Management
0.3
Revenue
Management
0.45
People
Management
0.88
Management
Management
0.7
Culture
Management
0.95
Capacity
Management
0.5
Experience
Management
0.68
Innovation
Management
0.8
APPENDIX E. FIELD RESEARCH DATA
298
From
To
Weight
Brand
Management
0.55
Alliances
Management
0.35
Safety
Management
0.45
Distribution
Management
0.4
Optimization
Management
0.85
Information
Management
0.73
Customer-centric
Management
0.8
Network
Culture
0.45
Revenue
Culture
0.43
People
Culture
0.98
Management
Culture
0.98
Culture
Culture
0.6
Capacity
Culture
0.35
Experience
Culture
0.75
Innovation
Culture
0.9
Brand
Culture
0.7
Alliances
Culture
0.4
Safety
Culture
0.5
Distribution
Culture
0.33
Optimization
Culture
0.75
Information
Culture
0.68
Customer-centric
Culture
0.88
Network
Capacity
0.98
Revenue
Capacity
0.93
People
Capacity
0.6
Management
Capacity
0.65
Culture
Capacity
0.35
APPENDIX E. FIELD RESEARCH DATA
299
From
To
Weight
Capacity
Capacity
0.8
Experience
Capacity
0.23
Innovation
Capacity
0.43
Brand
Capacity
0.45
Alliances
Capacity
0.63
Safety
Capacity
0.4
Distribution
Capacity
0.55
Optimization
Capacity
0.68
Information
Capacity
0.63
Customer-centric
Capacity
0.5
Network
Experience
0.75
Revenue
Experience
0.43
People
Experience
0.88
Management
Experience
0.85
Culture
Experience
0.8
Capacity
Experience
0.65
Experience
Experience
0.7
Innovation
Experience
0.8
Brand
Experience
0.68
Alliances
Experience
0.53
Safety
Experience
0.38
Distribution
Experience
0.75
Optimization
Experience
0.63
Information
Experience
0.78
Customer-centric
Experience
1
Network
Innovation
0.43
Revenue
Innovation
0.33
APPENDIX E. FIELD RESEARCH DATA
300
From
To
Weight
People
Innovation
0.9
Management
Innovation
0.98
Culture
Innovation
0.95
Capacity
Innovation
0.25
Experience
Innovation
0.65
Innovation
Innovation
0.48
Brand
Innovation
0.53
Alliances
Innovation
0.43
Safety
Innovation
0.33
Distribution
Innovation
0.55
Optimization
Innovation
0.53
Information
Innovation
0.73
Customer-centric
Innovation
0.78
Network
Brand
0.6
Revenue
Brand
0.35
People
Brand
0.8
Management
Brand
0.83
Culture
Brand
0.85
Capacity
Brand
0.45
Experience
Brand
0.7
Innovation
Brand
0.85
Brand
Brand
0.6
Alliances
Brand
0.63
Safety
Brand
0.53
Distribution
Brand
0.43
Optimization
Brand
0.58
Information
Brand
0.58
APPENDIX E. FIELD RESEARCH DATA
301
From
To
Weight
Customer-centric
Brand
0.9
Network
Alliances
0.78
Revenue
Alliances
0.65
People
Alliances
0.45
Management
Alliances
0.7
Culture
Alliances
0.43
Capacity
Alliances
0.45
Experience
Alliances
0.55
Innovation
Alliances
0.48
Brand
Alliances
0.65
Alliances
Alliances
0.6
Safety
Alliances
0.2
Distribution
Alliances
0.68
Optimization
Alliances
0.53
Information
Alliances
0.7
Customer-centric
Alliances
0.65
Network
Safety
0.2
Revenue
Safety
0.15
People
Safety
0.83
Management
Safety
0.83
Culture
Safety
0.85
Capacity
Safety
0.33
Experience
Safety
0.3
Innovation
Safety
0.6
Brand
Safety
0.48
Alliances
Safety
0.35
Safety
Safety
0.7
APPENDIX E. FIELD RESEARCH DATA
302
From
To
Weight
Distribution
Safety
0.15
Optimization
Safety
0.68
Information
Safety
0.5
Customer-centric
Safety
0.28
Network
Distribution
0.8
Revenue
Distribution
0.93
People
Distribution
0.33
Management
Distribution
0.55
Culture
Distribution
0.43
Capacity
Distribution
0.55
Experience
Distribution
0.5
Innovation
Distribution
0.73
Brand
Distribution
0.55
Alliances
Distribution
0.53
Safety
Distribution
0.2
Distribution
Distribution
0.6
Optimization
Distribution
0.73
Information
Distribution
0.73
Customer-centric
Distribution
0.88
Network
Optimization
0.65
Revenue
Optimization
0.6
People
Optimization
0.68
Management
Optimization
0.73
Culture
Optimization
0.88
Capacity
Optimization
0.55
Experience
Optimization
-0.68
Innovation
Optimization
0.75
APPENDIX E. FIELD RESEARCH DATA
303
From
To
Weight
Brand
Optimization
0.25
Alliances
Optimization
0.45
Safety
Optimization
0.55
Distribution
Optimization
0.5
Optimization
Optimization
0.7
Information
Optimization
0.7
Customer-centric
Optimization
0.53
Network
Information
0.48
Revenue
Information
0.65
People
Information
0.5
Management
Information
0.5
Culture
Information
0.53
Capacity
Information
0.3
Experience
Information
0.58
Innovation
Information
0.58
Brand
Information
0.18
Alliances
Information
0.43
Safety
Information
0.3
Distribution
Information
0.6
Optimization
Information
0.58
Information
Information
0.7
Customer-centric
Information
0.73
Network
Customer-centric
0.7
Revenue
Customer-centric
0.55
People
Customer-centric
0.83
Management
Customer-centric
0.85
Culture
Customer-centric
0.93
APPENDIX E. FIELD RESEARCH DATA
304
From
To
Weight
Capacity
Customer-centric
0.3
Experience
Customer-centric
0.98
Innovation
Customer-centric
0.9
Brand
Customer-centric
0.73
Alliances
Customer-centric
0.55
Safety
Customer-centric
0.38
Distribution
Customer-centric
0.63
Optimization
Customer-centric
0.53
Information
Customer-centric
0.65
Customer-centric
Customer-centric
0.7
Network
OpMargin
0.93
Revenue
OpMargin
0.95
People
OpMargin
0.9
Management
OpMargin
0.93
Culture
OpMargin
0.75
Capacity
OpMargin
0.93
Experience
OpMargin
0.8
Innovation
OpMargin
0.85
Brand
OpMargin
0.68
Alliances
OpMargin
0.63
Safety
OpMargin
0.43
Distribution
OpMargin
0.78
Optimization
OpMargin
0.88
Information
OpMargin
0.78
Customer-centric
OpMargin
0.88
APPENDIX E. FIELD RESEARCH DATA
305
E.9 Vertices Attributes (vertices.csv)
name
label
type
id
Regulation
RE
constraint
1
Fuel
FU
constraint
2
Competition
CP
constraint
3
Commoditization
CO
constraint
4
Unions
UN
constraint
5
Labor
LA
constraint
6
Slots
SL
constraint
7
ExCapacity
EC
constraint
8
Capital
CL
constraint
9
Biz-demand
BZ
constraint
10
Network
NW
valuerepo
11
Revenue
RV
valuerepo
12
People
PE
valuerepo
13
Management
MM
valuerepo
14
Culture
CU
valuerepo
15
Capacity
CM
valuerepo
16
Experience
EX
valuerepo
17
Innovation
IV
valuerepo
18
Brand
BD
valuerepo
19
Alliances
AL
valuerepo
20
Safety
SF
valuerepo
21
Distribution
DI
valuerepo
22
Optimization
OP
valuerepo
23
Information
IN
valuerepo
24
APPENDIX E. FIELD RESEARCH DATA
306
name
label
type
id
Customer-centric
CC
valuerepo
25
OpMargin
OM
om
26
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
307
APPENDIX F.
R CODE FOR REPRODUCIBLE RESEARCH
F.1 Code for Delphi Method Analysis
F.1.1 Required R libraries
library(dplyr)
library(plyr)
library(xlsx)
library(gtools)
library(ggplot2)
library(vcd)
library(vcdExtra)
library(grid)
library(gridExtra)
F.1.2 Analysis of Panel members
# load and read participants raw data
data_profile_raw <- read.xlsx("participants.xlsx", sheetIndex= 1, heade
r = TRUE)
data_participation_raw <- read.xlsx("participants.xlsx", sheetIndex= 2,
header = TRUE)
# summary of participants profiles
profile <- count(data_profile_raw, "Area")
profile_tot <- data.frame(Area= "Total", freq= sum(profile$freq) )
rbind(profile, profile_tot)
# summary of participants per geography and Delphi round
geo <- table(data_participation_raw$Geography[1:63], data_participation
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
308
_raw$Round[1:63])
class(geo)
## [1] "table"
# total number of valid responses received in Delphi R1:R4
geo_tot <- sum(sum(geo))
geo_tot
## [1] 63
# total number of invalid responses received in Delphi R1:R4
geo_tot_in <- count(data_participation_raw[64:91,2])
geo_tot_in
# plot of participants per geograghy and Delphi round
barplot(geo, main="Geography of Participants",
xlab="Delphi rounds", col=c("darkblue", "darkgreen", "red", "or
ange"),
legend = c("Europe", "Middle East", "North America", "South A
merica"))
# create participants by positition data frame
position_raw <- count(data_profile_raw$Position)
names(position_raw) <- c("Position", "Frequency")
position_order <- position_raw[order(position_raw$Frequency, decreasing
= TRUE),]
# add a "Total" row
position_tot <- data.frame(Position= "Total", Frequency= sum(position_o
rder$Frequency))
position_order <- rbind(position_order, position_tot)
# print participants by position data frame
position_order
# summary of type of responses per Delphi round R1:R4
responses_type <- table(data_participation_raw$Type, data_participation
_raw$Round)
# plot of participants per geograghy and Delphi round
barplot(responses_type, main="Type of Responses",
xlab="Delphi rounds", col=c("darkblue", "darkgreen", "red", "or
ange"),
legend = rownames(responses_type))
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
309
F.1.3 Delphi round 1 analysis
# load and read raw data file
r1_constraints_raw <- read.xlsx("round_1_R.xlsx", sheetIndex= 1, header
= TRUE)
r1_valuerepos_raw <- read.xlsx("round_1_R.xlsx", sheetIndex = 2, header
= TRUE)
# Summary of Categories by Variable
r1_categories <- data.frame("Constraints"= nrow(r1_constraints_raw), "V
alue Repositories"= nrow(r1_valuerepos_raw))
# Constraints, summary of top 10 categories
# add column "Percentage" showing the percentage of relative frequencie
s
r1_constraints_percent <- round((r1_constraints_raw$Frequency1/sum(r1_c
onstraints_raw$Frequency1))*100, digits=0)
r1_constraints_raw$Percentage <- r1_constraints_percent
## subset data frame "constraints_r1_raw" to the show only the first 10
categories
r1_constraints_summary <- r1_constraints_raw[1:10,]
## add column "Other" to show a summary of categories 11:39
other <- data.frame(Constraint1= "Other constraints", Frequency1= sum(r
1_constraints_raw$Frequency1[11:39]), Percentage= round(sum(r1_constrai
nts_raw$Frequency1[11:39])/sum(r1_constraints_raw$Frequency1)*100, digi
ts=0))
r1_constraints_other <- rbind(r1_constraints_summary, other)
## add column "Total"
total <- data.frame(Constraint1= "TOTAL", Frequency1= sum(r1_constraint
s_other$Frequency1[1:11]), Percentage= sum(r1_constraints_other$Percent
age[1:11]))
r1_constraints_final <- rbind(r1_constraints_other, total)
## print sumnmary table of frequencies
r1_constraints_final
# Value Repositories, summary of top 15 categories
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
310
## add column "Percentage" showing the percentage of relative requencie
s
r1_valuerepos_percent <- round((r1_valuerepos_raw$Frequency1/sum(r1_val
uerepos_raw$Frequency1))*100, digits= 2)
r1_valuerepos_raw$Percentage <- r1_valuerepos_percent
## subset data frame "valuerepos_r1_raw" to the show only the first 15
categories
r1_valuerepos_summary <- r1_valuerepos_raw[1:15,]
## add column "Other" to show a summary of categories 16:29
other <- data.frame(ValueRepository1= "Other Value repositories", Frequ
ency1= sum(r1_valuerepos_raw$Frequency1[16:29]), Percentage= round(sum(
r1_valuerepos_raw$Frequency1[16:29])/sum(r1_valuerepos_raw$Frequency1)*
100, digits=2))
r1_valuerepos_other <- rbind(r1_valuerepos_summary, other)
## add column "Total"
total <- data.frame(ValueRepository1= "TOTAL", Frequency1= sum(r1_value
repos_other$Frequency1[1:16]), Percentage= sum(r1_valuerepos_other$Perc
entage[1:16]))
r1_valuerepos_final <- rbind(r1_valuerepos_other, total)
## print sumnmary table of frequencies
r1_valuerepos_final
F.1.4 Delphi round 2 analysis
# load and read raw data file
r2_constraints_raw <- read.xlsx("round_2_R.xlsx", sheetIndex= 1, header
= TRUE)
r2_valuerepos_raw <- read.xlsx("round_2_R.xlsx", sheetIndex = 2, header
= TRUE)
# Constraints, summary of top 10 categories
# add column "Percentage" showing the percentage of relative frequencie
s
r2_constraints_percent <- round((r2_constraints_raw$Adj.Frequency/sum(r
2_constraints_raw$Adj.Frequency))*100, digits=0)
r2_constraints_raw$Percentage <- r2_constraints_percent
## subset data frame "constraints_r2_raw" to the show only the first 10
categories
r2_constraints_summary10 <- r2_constraints_raw[1:10,c(1,2,4)]
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
311
## add column "Other" to show a summary of categories 11:33
other <- data.frame(Constraint2= "Other constraints", Frequency2= sum(r
2_constraints_raw$Frequency2[11:33]), Percentage= round(sum(r2_constrai
nts_raw$Adj.Frequency[11:33])/sum(r2_constraints_raw$Adj.Frequency)*100
, digits=0))
r2_constraints_other <- rbind(r2_constraints_summary10, other)
## add column "Total"
total <- data.frame(Constraint2= "TOTAL", Frequency2= sum(r2_constraint
s_other$Frequency2[1:11]), Percentage= sum(r2_constraints_other$Percent
age[1:11]))
r2_constraints_final <- rbind(r2_constraints_other, total)
## print sumnmary table of frequencies
r2_constraints_final
# Value Repositories, summary of top 10 categories
## add column "Percentage" showing the percentage of relative frequenci
es
r2_valuerepos_percent <- round((r2_valuerepos_raw$Adj.Frequency/sum(r2_
valuerepos_raw$Adj.Frequency))*100, digits=0)
r2_valuerepos_raw$Percentage <- r2_valuerepos_percent
## subset data frame "constraints_r2_raw" to the show only the first 10
categories
r2_valuerepos_summary10 <- r2_valuerepos_raw[1:10,c(1,2,4)]
r2_valuerepos_summary10
# add column "Other" to show a summary of categories 11:33
other <- data.frame(ValueRepository2= "Other Value Repositories", Frequ
ency2= sum(r2_valuerepos_raw$Frequency2[11:29]), Percentage= round(sum(
r2_valuerepos_raw$Adj.Frequency[11:29])/sum(r2_valuerepos_raw$Adj.Frequ
ency)*100, digits=0))
r2_valuerepos_other <- rbind(r2_valuerepos_summary10, other)
## add column "Total"
total <- data.frame(ValueRepository2= "TOTAL", Frequency2= sum(r2_value
repos_other$Frequency2[1:11]), Percentage= sum(r2_valuerepos_other$Perc
entage[1:11]))
r2_valuerepos_final <- rbind(r2_valuerepos_other, total)
## print sumnmary table of frequencies
r2_valuerepos_final
# Comparison of "Constraints" vs Value Repositories in Delphi R1:R2
constraints_merge <- merge(r1_constraints_raw, r2_constraints_raw, by.x
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
312
="Constraint1", by.y= "Constraint2", all= TRUE)
constraints_merge_reduc1 <- constraints_merge[order(constraints_merge$A
dj.Frequency, decreasing = TRUE),]
constraints_merge_reduc2 <- constraints_merge_reduc1[,c(1,2,4)]
constraints_merge_reduc2[1:10,]
# comparison of "Value_repositories" in Delphi R1:R2
valuerepos_merge <- merge(r1_valuerepos_raw, r2_valuerepos_raw, by.x="V
alueRepository1", by.y= "ValueRepository2", all= TRUE)
valuerepos_merge_reduc1 <- valuerepos_merge[,c(1,2,4)]
valuerepos_merge_reduc2 <- valuerepos_merge_reduc1[order(valuerepos_mer
ge_reduc1$Frequency2, decreasing = TRUE),]
names(valuerepos_merge_reduc2) <- c("ValueRepository", "Frequency R1",
"Frequency R2")
valuerepos_merge_reduc2[1:15,]
F.1.5 Delphi round 3 analysis
# load Constraints vs Value Repositories raw data
r3_cons_raw <- read.csv("round_3_cons_R.csv", header= TRUE)
# make Strength an ordered variable
r3_cons_raw$Strength <- ordered(r3_cons_raw$Strength, levels=c("Z","VW"
, "W","S","VS"))
# load Value Repositories vs Value Repositories raw data
r3_vr_raw <- read.csv("round_3_vr_R.csv", header= TRUE)
# make Strength an ordered variable
r3_vr_raw$Strength <- ordered(r3_vr_raw$Strength, levels=c("Z","VW", "W
","S","VS"))
# load Value Repositories vs Operating Margin raw data
r3_om_raw <- read.csv("round_3_om_R.csv", header= TRUE)
# make Strength an ordered variable
r3_om_raw$Strength <- ordered(r3_om_raw$Strength, levels=c("Z","VW", "W
","S","VS"))
# plot Constraints vs Value Repositories Heatmap
r3_cons_heat<- r3_cons_raw %>%
group_by(Constraint, ValueRepository) %>%
slice(which.max(Freq))
heat.cons3 <- ggplot(r3_cons_heat, cex = 0.5, aes(x = ValueRepository,
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
313
y= Constraint, fill = Strength)) + ggtitle("Round 3: Constraints vs Val
ue Repositories") + geom_tile() + scale_fill_manual(values = c("#F2F2F2
","#F22738","#EB91B4","#01ABE9","#1B346C")) + ylim(rev(levels(r3_cons_h
eat$Constraint)))
heat.cons3 + theme(axis.text.x = element_text(angle = 45, hjust = 1, si
ze=18), axis.text.y = element_text(size=18), legend.text=element_text(s
ize=14), plot.title = element_text(vjust=1, face="bold"))
# plot Value Repositories vs Value Repositories Heatmap
r3_vr_heat<- r3_vr_raw %>%
group_by(ValueRepository1, ValueRepository2) %>%
slice(which.max(Freq))
heat.vr3 <- ggplot(r3_vr_heat, aes(x = ValueRepository1, y= ValueReposi
tory2, fill = Strength)) + ggtitle("Round 3: Value Repositories vs Valu
e Repositories") + geom_tile() + scale_fill_manual(values = c("#F2F2F2"
,"#F22738","#EB91B4","#01ABE9","#1B346C")) + ylim(rev(levels(r3_vr_heat
$ValueRepository2)))
heat.vr3 + theme(axis.text.x = element_text(angle = 45, hjust = 1, size
=18), axis.text.y = element_text(size=18), legend.text=element_text(siz
e=14), plot.title = element_text(vjust=1, face="bold"))
# plot Value Repositories vs Operating Margin
# convert data frame to table
r3_om.tab <- xtabs(Freq~ValueRepository+Strength, data=r3_om_raw)
# cell percentages
r3_om.tab.prop <- prop.table(r3_om.tab,1)
# plot a mosaic
mosaic(r3_om.tab.prop, gp = gpar(fill=c("red","orange","light green","g
reen","blue"), fontsize=12), main="Value Repositories vs Operating Marg
in (%)", labeling_args = list(rot_labels = c(bottom = 90,left = 0), off
set_varnames = c(left = 5), offset_labels = c(left = 2.5),margins = c(l
eft = 4, bottom = 3)))
F.1.6 Delphi round 4 analysis
# load Constraints raw data- STRENGTH
r4_consst_raw <- read.csv("round_4_consst_R.csv", header= TRUE)
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
314
# make Strength an ordered variable
r4_consst_raw$Strength <- ordered(r4_consst_raw$Strength, levels=c("Z",
"VW", "W","S","VS"))
# load Constraints raw data- SIGN
r4_conssi_raw <- read.csv("round_4_conssi_R.csv", header= TRUE)
# make Sign an ordered variable
r4_conssi_raw$Sign <- ordered(r4_conssi_raw$Sign, levels=c("Neg","Neu",
"Pos"))
# load Value Repositories raw data- STRENGTH
r4_vrst_raw <- read.csv("round_4_vrst_R.csv", header= TRUE)
# make Strength an ordered variable
r4_vrst_raw$Strength <- ordered(r4_vrst_raw$Strength, levels=c("Z","VW"
, "W","S","VS"))
# load Value Repositories raw data- SIGN
r4_vrsi_raw <- read.csv("round_4_vrsi_R.csv", header= TRUE)
# make Sign an ordered variable
r4_vrsi_raw$Sign <- ordered(r4_vrsi_raw$Sign, levels=c("Neg","Neu", "Po
s"))
# load Operating Margin raw data- STRENGTH
r4_omst_raw <- read.csv("round_4_omst_R.csv", header= TRUE)
# make Strength an ordered variable
r4_omst_raw$Strength <- ordered(r4_omst_raw$Strength, levels=c("Z","VW"
, "W","S","VS"))
# load Operating Margin raw data- SIGN
r4_omsi_raw <- read.csv("round_4_omsi_R.csv", header= TRUE)
# make Sign an ordered variable
r4_omsi_raw$Sign <- ordered(r4_omsi_raw$Sign, levels=c("Neg","Neu", "Po
s"))
# plot Constraints vs Value Repositories Heatmap- STRENGTH
r4_consst_heat<- r4_consst_raw %>%
group_by(Constraint, ValueRepository) %>%
slice(which.max(Freq))
heat.consst4 <- ggplot(r4_consst_heat, aes(x = ValueRepository, y= Cons
traint, fill = Strength)) + ggtitle("Round 4: Constraints vs Value Repo
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
315
sitories- STRENGTH") + geom_tile() + scale_fill_manual(values = c("#F2F
2F2","#F22738","#EB91B4","#01ABE9","#1B346C")) + ylim(rev(levels(r4_con
sst_heat$Constraint)))
heat.consst4 + theme(axis.text.x = element_text(angle = 45, hjust = 1,
size=18), axis.text.y = element_text(size=18), legend.text=element_text
(size=14), plot.title = element_text(vjust=1, face="bold"))
# plot Value Repositories vs Value Repositories Heatmap- STRENGTH
r4_vrst_heat<- r4_vrst_raw %>%
group_by(ValueRepository1, ValueRepository2) %>%
slice(which.max(Freq))
heat.vr4 <- ggplot(r4_vrst_heat, aes(x = ValueRepository1, y= ValueRepo
sitory2, fill = Strength)) + ggtitle("Round 4: Value Repositories vs Va
lue Repositories- STRENGTH") + geom_tile() + scale_fill_manual(values =
c("#F2F2F2","#F22738","#EB91B4","#01ABE9","#1B346C")) + ylim(rev(levels
(r4_vrst_heat$ValueRepository2)))
heat.vr4 + theme(axis.text.x = element_text(angle = 45, hjust = 1, size
=18), axis.text.y = element_text(size=18), legend.text=element_text(siz
e=14), plot.title = element_text(vjust=1, face="bold"))
# plot Value Repositories vs Operating Margin- STRENGTH
r4_omst.tab <- xtabs(Freq~ValueRepository+Strength, data=r4_omst_raw)
# cell percentages
r4_omst.tab.prop <- prop.table(r4_omst.tab,1)
# plot a mosaic
mosaic(r4_omst.tab.prop, gp = gpar(fill=c("red","orange","light green",
"green","blue"), fontsize=12), main="Value Repositories vs Operating Ma
rgin- STRENGTH (%)", labeling_args = list(rot_labels = c(bottom = 90,le
ft = 0), offset_varnames = c(left = 5), offset_labels = c(left = 2.5),m
argins = c(left = 4, bottom = 3)))
F.2 Code for Network Analysis
F.2.1 Required R libraries
library(igraph)
library(sna)
library(network)
library(ape)
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
316
F.2.2 Eigen Centrality
# load data
edgespanel <- read.csv("edgespanel.csv", header= TRUE, stringsAsFactors
= FALSE)
vertices <- read.csv("vertices.csv", header= TRUE, stringsAsFactors = F
ALSE)
# make edges weights as numeric
edgespanel$weight <- as.numeric(as.character(sub("," , ".", edgespanel$
weight)))
# make a data frame with the edges and vertices attributes
paneldata.g <- graph.data.frame(edgespanel[151:375,], directed= "TRUE",
vertices= vertices[11:25,])
# eigen_centrality
eigen_centrality(paneldata.g, directed= TRUE, scale= TRUE, weights = NU
LL)
F.2.3 Hierarchical clustering
# load data
edgesraw <- read.csv("edgespanel.csv", header= TRUE, stringsAsFactors =
FALSE)
verticesraw <- read.csv("vertices.csv", header= TRUE, stringsAsFactors
= FALSE)
edgespanel <- edgesraw[151:375,]
vertices <- verticesraw[11:25,]
# make edges weights as numeric
edgespanel$weight <- as.numeric(as.character(sub("," , ".", edgespanel$
weight)))
# make a data frame with the edges and vertices attributes
paneldata.g <- graph.data.frame(edgespanel, directed= "TRUE", vertices=
vertices)
# delete edges with weight<0.5
panel.g <- delete_edges(paneldata.g, which(E(paneldata.g)$weight <=0.75
))
# Fast agglomerative hierarchical clustering algorithm
panel.und <- as.undirected(panel.g)
panel.agc <- fastgreedy.community(panel.und)
length(panel.agc)
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
317
## [1] 2
sizes(panel.agc)
## Community sizes
## 1 2
## 6 9
membership(panel.agc)
## Network Revenue People Management
## 1 1 2 2
## Culture Capacity Experience Innovation
## 2 1 2 2
## Brand Alliances Safety Distribution
## 2 1 2 1
## Optimization Information Customer-centric
## 2 1 2
plot(panel.agc, panel.g, edge.arrow.size=0.2)
# Corresponding dendrogram for this partitioning
dendPlot(panel.agc, mode= "phylo")
F.2.4 Topological network complexity methods
#########################################
# Basic Network Metrics
#########################################
is.simple(panel.g)
## [1] FALSE
igraph::is.connected(panel.g)
## [1] TRUE
igraph::is.connected(panel.g, "weak") # see clusters
## [1] TRUE
igraph::is.connected(panel.g, "strong") # see clusters
## [1] FALSE
diameter(panel.g, directed = TRUE)
## [1] 1.93
########################################
# Vertex Degree
########################################
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
318
# Group of vertices by degree
hist(degree(panel.m), col="lightblue", xlim=c(0,40), xlab="Vertex Degre
e")
sort(degree(panel.m, "all"), decreasing= TRUE)
## [1] 23.37 21.73 21.17 19.91 19.48 19.44 18.95 18.83 17.88 17.62 17.
43
## [12] 16.73 16.64 15.31 11.67 11.30 7.29 7.25 6.82 4.87 4.35 3.
38
## [23] 2.88 2.81 2.46 1.93
sort(degree(panel.m, "total"), decreasing= TRUE)
## [1] 23.37 21.73 21.17 19.91 19.48 19.44 18.95 18.83 17.88 17.62 17.
43
## [12] 16.73 16.64 15.31 11.67 11.30 7.29 7.25 6.82 4.87 4.35 3.
38
## [23] 2.88 2.81 2.46 1.93
sort(degree(panel.m, "out"), decreasing= TRUE)
## [1] 23.37 21.73 21.17 19.91 19.48 19.44 18.95 18.83 17.88 17.62 17.
43
## [12] 16.73 16.64 15.31 11.67 11.30 7.29 7.25 6.82 4.87 4.35 3.
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
319
38
## [23] 2.88 2.81 2.46 1.93
sort(degree(panel.m, "in"), decreasing = TRUE)
## [1] 23.37 21.73 21.17 19.91 19.48 19.44 18.95 18.83 17.88 17.62 17.
43
## [12] 16.73 16.64 15.31 11.67 11.30 7.29 7.25 6.82 4.87 4.35 3.
38
## [23] 2.88 2.81 2.46 1.93
mean(igraph::degree(panel.g))
## [1] 18.69231
# Vertex strength: sum of weights of edges incident to a given vertex
hist(graph.strength(panel.g), col="pink", xlab="Vertex Strength")
########################################
# Vertex Centrality
########################################
# Number of clusters found
panel.scc <- clusters(panel.g, mode=c("strong"))
table(panel.scc$size)
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
320
## < table of extent 0 >
# Number of neighbors per a concrete vertex
neighbors(panel.g,5)
## + 4/26 vertices, named:
## [1] People Management Culture Optimization
# Vertex centrality by degree
central.panel.m <- network::as.network.matrix(panel.m)
gplot.target(central.panel.m, degree(central.panel.m), main="Degree", c
irc.lab= FALSE, circ.col="skyblue", usearrows = FALSE, vertex.col=c("bl
ue", rep("red", 32), "yellow"), edge.col="darkgray")
# Vertex centrality by closeness
central.panel.m <- network::as.network.matrix(panel.m)
gplot.target(central.panel.m, closeness(central.panel.m), main="Closene
ss", circ.lab= FALSE, circ.col="skyblue", usearrows = FALSE, vertex.col
=c("blue", rep("red", 32), "yellow"), edge.col="darkgray")
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
321
# Vertex centrality by betweenness
central.panel.m <- network::as.network.matrix(panel.m)
gplot.target(central.panel.m, betweenness(central.panel.m), main="Betwe
enness", circ.lab= FALSE, circ.col="skyblue", usearrows = FALSE, vertex
.col=c("blue", rep("red", 32), "yellow"), edge.col="darkgray")
# Vertex centrality by eigenvector
central.panel.m <- network::as.network.matrix(panel.m)
gplot.target(central.panel.m, evcent(central.panel.m), main="Eigen Vect
or", circ.lab= FALSE, circ.col="skyblue", usearrows = FALSE, vertex.col
=c("blue", rep("red", 32), "yellow"), edge.col="darkgray")
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
322
# Hubs and Authorities
l <- layout.kamada.kawai(panel.g)
plot(panel.g, layout=l, edge.arrow.size=0.2, main="Hubs", vertex.label=
"", vertex.size=10 * sqrt(hub.score(panel.g)$vector))
plot(panel.g, layout=l, edge.arrow.size=0.2, main="Authorities", vertex
.label="", vertex.size=10 *sqrt(authority.score(panel.g)$vector))
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
323
########################################
# Edges characteristics
########################################
# Edge betweennnes centrality
edgebtw <- edge.betweenness(panel.g)
E(panel.g)[order(edgebtw, decreasing= T)[1:10]]
## + 10/243 edges (vertex names):
## [1] Slots ->Alliances Capacity ->Optimization
## [3] Capital ->Optimization Capacity ->Management
## [5] Alliances ->Experience Optimization->Brand
## [7] Unions ->People Labor ->Optimization
## [9] Optimization->Culture Safety ->Brand
########################################
# Network cohesion
########################################
# Census of cliques
table(sapply(cliques(panel.g), length)) # how structured the graph is
## 1 2 3 4 5 6 7 8 9 10
## 26 168 568 1234 1813 1803 1190 497 118 12
cliques(panel.g)[sapply(cliques(panel.g), length) == 10]
## [[1]]
## + 10/26 vertices, named:
## [1] Competition Network Revenue Management
## [5] Capacity Alliances Distribution Optimization
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
324
## [9] Information Customer-centric
##
## [[2]]
## + 10/26 vertices, named:
## [1] Competition Network Revenue Management
## [5] Brand Alliances Distribution Optimization
## [9] Information Customer-centric
##
## [[3]]
## + 10/26 vertices, named:
## [1] Competition Network Management Capacity
## [5] Experience Alliances Distribution Optimization
## [9] Information Customer-centric
##
## [[4]]
## + 10/26 vertices, named:
## [1] Competition Network Management Experience
## [5] Brand Alliances Distribution Optimization
## [9] Information Customer-centric
##
## [[5]]
## + 10/26 vertices, named:
## [1] Competition Revenue People Management
## [5] Culture Innovation Brand Optimization
## [9] Information Customer-centric
##
## [[6]]
## + 10/26 vertices, named:
## [1] Competition People Management Culture
## [5] Experience Innovation Brand Optimization
## [9] Information Customer-centric
##
## [[7]]
## + 10/26 vertices, named:
## [1] Network Revenue Management Capacity
## [5] Alliances Distribution Optimization Information
## [9] Customer-centric OpMargin
##
## [[8]]
## + 10/26 vertices, named:
## [1] Network Revenue Management Brand
## [5] Alliances Distribution Optimization Information
## [9] Customer-centric OpMargin
##
## [[9]]
## + 10/26 vertices, named:
## [1] Network Management Capacity Experience
## [5] Alliances Distribution Optimization Information
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
325
## [9] Customer-centric OpMargin
##
## [[10]]
## + 10/26 vertices, named:
## [1] Network Management Experience Brand
## [5] Alliances Distribution Optimization Information
## [9] Customer-centric OpMargin
##
## [[11]]
## + 10/26 vertices, named:
## [1] Revenue People Management Culture
## [5] Innovation Brand Optimization Information
## [9] Customer-centric OpMargin
##
## [[12]]
## + 10/26 vertices, named:
## [1] People Management Culture Experience
## [5] Innovation Brand Optimization Information
## [9] Customer-centric OpMargin
clique.number(panel.g) # the size of the largest clique
## Warning in .Call("R_igraph_clique_number", graph, PACKAGE = "igraph"
): At
## cliques.c:908 :directionality of edges is ignored for directed graph
s
## [1] 10
# Maximal cliques
table(sapply(maximal.cliques(panel.g), length)) # a clique that is not
a subset of a larger clique
## Warning in .Call("R_igraph_maximal_cliques", graph, subset,
## as.numeric(min), : At maximal_cliques_template.h:203 :Edge direction
s are
## ignored for maximal clique calculation
##
## 3 4 5 6 7 8 9 10
## 1 5 3 3 5 5 14 12
# k-core decomposition of the network
panelcores <- graph.coreness(panel.g) #coreness one(black), two(red),
three(green), four(blue)
gplot.target(panel.m, panelcores, circ.lab = FALSE, circ.col="skyblue",
usearrows = FALSE, vertex.col=panelcores, edge.col="darkgray")
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
326
# Census of dyads and triads
panel.g<- simplify(panel.g)
dyad_census(panel.g) # match this analysis with that of hubs and author
ities
## $mut
## [1] 61
##
## $asym
## [1] 107
##
## $null
## [1] 157
# Motifs
graph.motifs(panel.g, size=3) #small connected subgraphs of interest
## [1] NA NA 238 NA 124 238 46 73 182 53 47 0 17 89 92 115
# Transitivity
transitivity(panel.g) #measure of global clustering, relative frequency
with which connected triples close to form triangles
## [1] 0.6955102
# Reciprocity
reciprocity(panel.g, mode="default") # match this result with that of d
yad census
## [1] 0.5327511
reciprocity(panel.g, mode="ratio")
## [1] 0.3630952
########################################
# Connectivity
########################################
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
327
# Census of all connected components within the graph
comps <- decompose.graph(panel.g) # see giant components
table(sapply(comps, vcount))
##
## 26
## 1
table(sapply(decompose.graph(panel.g), vcount)) # alternative code
##
## 26
## 1
# Further analysis of giant component, check for small world properties
panel.gc <- decompose.graph(panel.g)[[1]]
average.path.length(panel.gc) # the shortest path distance between pair
s of vertices is quite small
## [1] 1.420779
diameter(panel.gc) # the clustering is relatively high, check for i.e.
transitivity
## [1] 1.93
# Vertex and edge (high/low) connectivity
vertex.connectivity(panel.gc) # it requires the removal of .. vertex/ed
ge to break this subgraph into additional components
## [1] 0
edge.connectivity(panel.gc)
## [1] 0
# Indentify Vertex-cut or edge-cut, thus if the network is vulnerable
panel.cut.vertices <- articulation.points(panel.gc)
length(panel.cut.vertices)
## [1] 0
########################################
# Assortativity
########################################
# check whether edges only connect vertices of the same category (1), o
r no different from a random # assignment of edges
assortativity.nominal(panel.g, (V(panel.g)$type == "constraint")+1, dir
ected = TRUE)
## [1] 0
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
328
assortativity.nominal(panel.g, (V(panel.g)$type == "valuerepo")+1, dire
cted = TRUE)
## [1] -0.1125172
assortativity.nominal(panel.g, (V(panel.g)$type == "om")+1, directed =
TRUE)
## [1] 0
# degree-degree correlation of adjacent vertices
assortativity.degree(panel.g) # match this analysis with that of verte
x centrality
## [1] -0.2089687
F.3 Code for Fuzzy Cognitive Map
F.3.1 Required R libraries
# load libraries
library(igraph)
library(FCMapper)
F.3.2 Fuzzy Cognitive Map
# load and read data
edgespanel <- read.csv("edgespanel.csv", header= TRUE, stringsAsFactors
= FALSE)
vertices <- read.csv("vertices.csv", header= TRUE, stringsAsFactors = F
ALSE)
#edgespanel$weightws <- as.numeric(as.character(sub("," , ".", edgespan
el$weightws)))
# igraph data frame with the edges and vertices attributes
panel.g <- graph.data.frame(edgespanel, directed= "TRUE", vertices=vert
ices)
edgespanel.g <- delete_edges(panel.g, which(E(panel.g)$weightws <0.75))
# adjacency matrix
edgespanel.adj <- as_adjacency_matrix(edgespanel.g, attr = "weightws")
edgespanel.m <- as.matrix(edgespanel.adj)
diag(edgespanel.m) <- 1 #condition for Stylios Type II FCM
# save adjacency matrix to paste in document
# write.csv(edgespanel.m, file = "./FCM_adjmatrix.csv")
dimnames(edgespanel.m) <- NULL
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
329
# pass adj matrix to check.matrix
check.matrix(edgespanel.m)
## [1] "Matrix is square."
## [1] "All values of the matrix are within -1 and 1."
## [1] "The diagonal is not equal to 0 (ie, there is a self-loop). Cons
ider whether this is appropriate."
# set parameters for simulation scenario
concept.names <- vertices$name
###################################################
# Scenarios Settings
###################################################
iter = 30
lambda <- 0.2
w <- edgespanel.m
act_vector <- matrix(0, nrow = iter, ncol = length(w[1,]))
act_vector[1,] <- rep(0.5, 26)
# Scenario 1
#set.concepts <- c("Regulation", "Fuel", "Competition", "Labor", "Biz-d
emand")
#set.values <- c(-0.9, -0.9, -0.9, -0.9, -0.9)
#Scenario 2
#set.concepts = c("Optimization", "Innovation", "Experience", "Informat
ion", "Customer-centric")
#set.values = c(0.9, 0.9, 0.9, 0.9, 0.9)
#Scenario 3
#set.concepts = c("Network", "Revenue", "Capacity", "Distribution")
#set.values = c(0.9, 0.9, 0.9, 0.9)
#act_vector[1,which(concept.names %in% set.concepts == TRUE)] = set.val
ues
###################################################
# FCM Inference Engine with Hebbian Learning Rule
###################################################
alfa1 <- 0.1
#alfa2 <- 1
for (k in 2:iter) {
act_vector[k, ] = 1/(1 + exp(-lambda * (act_vector[k - 1,] %*% w)))
#w <- w + alfa1*(act_vector[i,]*act_vector[i-1,])-alfa2*(act_vector[i
,] + act_vector[i-1,])
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
330
#w <- w -alfa2*(act_vector[i,] + act_vector[i-1,])
for (i in 1:length(w[1,])) {
for (j in 1:length(w[1,])) {
if (i == j) {
dw <- 0
} else {
dci <- act_vector[i+1] - act_vector[i]
dcj <- act_vector[j+1] - act_vector[j]
if (dci * dcj > 0) {
dw <- dci * dcj
} else {
dw <- 0
}
}
w[i,j] <- w[i,j] + alfa1* dw
}
}
#act_vector[k, which(concept.names %in% set.concepts == TRUE)] = set.v
alues
#print(act_vector[k,])
}
# FCM outcome stability check
if (all.equal(act_vector[iter,], act_vector[iter-1,]) != TRUE) {
print("WARNING: Convergence not reached. Try increasing the number of
iterations.")
}
# FCM outcome data frame
results = data.frame(concept.names, act_vector[iter, ])
colnames(results) = c("Concept", "Equilibrium_value")
# plot results
plot(act_vector[, 1] ~ seq(1, iter, 1), type = "n", ylim = c(-0.15, 1),
xlab = "Time Step", ylab = "Value of AVCN component")
for (n in 1:length(w[1, ])) {
points(act_vector[, n] ~ seq(1, iter, 1), type = "l", col = n)
}
legend("bottom", legend = concept.names, cex = 0.7, col = seq(1, n, 1),
lty= 1, lwd=1.5, ncol=7, bty= "n", text.width=c(2.2,2.2,2.2,2.2))
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
331
F.4 Code for Panel Members Benchmarking Dotplots
# USER’S NOTE:
# Users need to change the letter "X" in the corresponding file names
# and variables in order to create the individual members’ dotplots
# upload required R libraries
library(lattice)
# load and read data
edgesX_raw <- read.csv("edgesX.csv", header= TRUE, stringsAsFactors = F
ALSE)
edgesX_raw$weight <- as.numeric(as.character(sub("," , ".", edgesX_raw$
weight)))
# creates dotplot Constraints vs Value Repositories
edgesX.cons <- edgesX_raw[c(1:150, 391:540),]
dotplot(constraint ~ weight | valuerepo, data= edgesX.cons, group = who
,
pch = c(1, 3), key = list(space = "right", transparent = TRUE,
points = list(pch = c(1, 3), col = 1:2), text = list(c("expert (you)",
"panel"))), scales=list(x=list(at=c(0, .25, .5, .75, 1), labels=c(0, .2
5, .5, .75, 1))),
panel = function(...) {
panel.superpose
panel.dotplot(...)
})
dev.copy(png,filename="dotplot_X_1.png", width = 1000, height = 650);
dev.off ()
# creates dotplot of Value Repositories
edgesX.vr <- edgesX_raw[c(151:375, 541:765),]
dotplot(constraint ~ weight | valuerepo, data= edgesX.vr, group = who,
pch = c(1, 3), key = list(space = "right", transparent = TRUE,
points = list(pch = c(1, 3), col = 1:2), text = list(c("expert (you)",
"panel"))), scales=list(x=list(at=c(0, .25, .5, .75, 1), labels=c(0, .2
5, .5, .75, 1))),
panel = function(...) {
panel.superpose
panel.dotplot(...)
})
dev.copy(png,filename="dotplot_X_2.png", width = 1000, height = 650);
dev.off ()
# creates dotplot of Value Repositories vs Operating Margin
edgesX.om <- edgesX_raw[c(376:390, 766:780),]
dotplot(constraint ~ weight | valuerepo, data= edgesX.om, group = who,
pch = c(1, 3), key = list(space = "right", transparent = TRUE,
points = list(pch = c(1, 3), col = 1:2), text = list(c("expert (you)",
APPENDIX F. R CODE FOR REPRODUCIBLE RESEARCH
332
"panel"))), scales=list(x=list(at=c(0, .25, .5, .75, 1), labels=c(0, .2
5, .5, .75, 1))),
panel = function(...) {
panel.superpose
panel.dotplot(...)
})
dev.copy(png,filename="dotplot_X_3.png", width = 1000, height = 650);
dev.off ()
APPENDIX G. VALUEINAIRLINES.COM
333
APPENDIX G.
VALUEINAIRLINES.COM
APPENDIX G. VALUEINAIRLINES.COM
334
APPENDIX G. VALUEINAIRLINES.COM
335
APPENDIX G. VALUEINAIRLINES.COM
336
APPENDIX G. VALUEINAIRLINES.COM
337
APPENDIX G. VALUEINAIRLINES.COM
338
APPENDIX G. VALUEINAIRLINES.COM
339
THIS PAGE INTENTIONALLY LEFT BLANK.
APPENDIX H. SOFTWARE TOOLS
341
APPENDIX H.
SOFTWARE TOOLS
H.1 R Programming Software
R is an open source software for statistical computation and graphics. The R distribution
includes the ability to save and run commands stored in script files, and an integrated editor in
the R Graphical User Interface (R-GUI). It compiles and runs on a wide variety of platforms
including Unix/Linux, PC, and Macintosh. Thousands of contributed packages are available, and
users are provided tools to make packages (Albert, Rizzo 2012).
At the core of R is an interpreted computer language. This language provides the logical
control of branching and looping, and modular programming using functions. The base R
distribution contains functions and data to implement and illustrate most common statistical
procedures, including linear and nonlinear modeling, classical statistical tests, time-series
analysis, classification, clustering analysis, density estimation, and much more.
One of R’s strengths is the ease with which well-designed publication-quality plots can
be produced, including mathematical symbols and formulae where needed. Great care has been
taken over the defaults for the minor design choices in graphics, but the user retains full control.
APPENDIX H. SOFTWARE TOOLS
342
An extensive suite of probability distribution functions and generators are provided, as
well as a graphical environment for exploratory data analysis and creating presentation
graphics. R is available as Free Software under the terms of the Free Software Foundation’s GNU
General Public License in source code form.
H.2 Network Workbench
Network Workbench (NWB) software, available at http://nwb.cns.iu.edu, is a large-scale
network analysis, modeling and visualization toolkit for biomedical, social science and physics
research. This software is used in the thesis to draw the AVCN graph and perform network
analysis.
The NWB supports network science research across scientific boundaries. Users of the
NWB can upload their own networks and perform network analysis with the most effective
algorithms available. In addition, they are able to generate, run, and validate network models to
advance their understanding of the structure and dynamics of particular networks. NWB also
provides advanced visualization tools to interactively explore and understand specific networks,
as well as their interaction with other types of networks.
The NWB provides members of the scientific research community at large with a direct
transfer of knowledge and results from the fields of specialist network research to a wider
scientific community. Researchers have access to validated algorithms that in the past have
been obtained through time-consuming personal developments of ad hoc computer programs
(Indiana, Northeastern et al. 2006).
APPENDIX H. SOFTWARE TOOLS
343
H.3 Qualtrics Survey System
Qualtrics (http://www.qualtrics.com) is an online survey management software that
enables users to do many kinds of online data collection and analysis including market research,
customer satisfaction and loyalty, product and concept testing, employee evaluations and
website feedback. According to recent news reports, Qualtrics has about 6,000 clients running
2.1 million surveys on an average day (Brustein 2014) This software is used in the thesis to
create and deliver the questionnaires of the field research Delphi process.
THIS PAGE INTENTIONALLY LEFT BLANK.
BIBLIOGRAPHY
345
BIBLIOGRAPHY
ADAMI, C., 2002. What is complexity. Bioessays, 24(12), pp. 1085-94.
ADLER, M. and ZIGLIO, E., 1996. Gazing into the oracle: The Delphi method and its
application to social policy and public health. Jessica Kingsley Publishers.
AHUJA, R.K., MAGNANTI, T.L. and ORLIN, J.B., 1993. Network flows: theory, algorithms,
and applications. Prentice Hall.
AHUJA, R.K., MAGNANTI, T.L., ORLIN, J.B. and REDDY, M.R., 1995. Applications of
network optimization. Handbooks in Operations Research and Management Science, 7, pp. 1-83.
AI-JUNKIE, , Kohonen's Self Organizing Feature Maps. Available: http://www.ai-
junkie.com/ann/som/som1.html.
ALBERGHINA, L. and WESTERHOFF, H.V., 2007. Systems biology: definitions and
perspectives. Springer Science & Business Media.
ALBERT, J. and RIZZO, M., 2012. R by Example. Springer Science & Business Media.
ALCHIAN, A.A. and DEMSETZ, H., 1972. Production, information costs, and economic
organization. The American Economic Review, , pp. 777-795.
ALLEE, V., 2002. A value network approach for modeling and measuring intangibles.
Madrid: Transparent Enterprise.
ALLEN, P., MAGUIRE, S. and MCKELVEY, B., 2011. The Sage Handbook of Complexity and
Management. Sage Publications.
AMMAN, H.M., TESFATSION, L., JUDD, K.L., KENDRICK, D.A. and RUST, J., 2006.
Handbook of Computational Economics. Elsevier.
AMOS, T. and PEARSE, N., 2008. Pragmatic research design: An illustration of the use of
the Delphi technique. Electronic Journal of Business Research Methods, 6(2), pp. 95-102.
ANDERSON, J.C., HAKANSSON, H. and JOHANSON, J., 1994. Dyadic business relationships
within a business network context. Journal of Marketing, 58(4), pp. 1.
BIBLIOGRAPHY
346
ANDERSON, N., HERRIOT, P. and HODGKINSON, G.P., 2001. The practitioner-researcher
divide in Industrial, Work and Organizational (IWO) psychology: Where are we now, and where
do we go from here? Journal of Occupational and Organizational Psychology, 74(4), pp. 391-411.
ANDO, T., 2010. Bayesian model selection and statistical modeling. CRC Press.
ARGENTI, J., 1997. Stakeholders: the case against. Long range planning, 30(3), pp. 442-
445.
ARROW, K.J., 1974. The Limits of Organization. New York, N.Y.: W. W. Norton and
Company.
ARROW, K.J., 1984. The Economics of Agency. Institute for Mathematical Studies in the
Social Sciences, Stanford University.
ARTHUR, W.B., 1999. Complexity and the economy. Science, (248), pp. 107-109.
ATIYA, A.F., 2001. Bankruptcy prediction for credit risk using neural networks: A survey
and new results. Neural Networks, IEEE Transactions on, 12(4), pp. 929-935.
AXTELL, R., 2000. Why agents? On the varied motivations for agent computing in the
social sciences. Retrieved from: http://www.brook.edu/es/dynamics/papers/agents/agents.pdf.
BAILEY, K.D., 1990. Social entropy theory. SUNY Press.
BALASUBRAMANIAN, R. and AGARWAL, D., 2013. Delphi Technique-A Review.
International Journal of Public Health Dentistry, 3(2), pp. 16-25.
BARABÁSI, A.L., 2002. Linked: The new science of networks. Perseus Pub.
BARABÁSI, A.L. and BONABEAU, E., 2003. Scale-free networks. Scientific American,
288(5), pp. 60-69.
BARBER, E., 2008. How to measure the "value" in value chains. International Journal of
Physical Distribution & Logistics Management, 38(9), pp. 685-698.
BARNEY, J., 1991. Firm resources and sustained competitive advantage. Journal of
management, 17(1), pp. 99-120.
BATTISTA, G.D., EADES, P., TAMASSIA, R. and TOLLIS, I.G., 1999. Graph drawing:
Algorithms for the visualisation of graphs. Prentice Hall.
BHOWMICK, K. and SHAH, M., 2015. Kohonen's Self-Organizing Feature Maps and Linear
Vector Quantization: A Comparison. International Journal of Computer Applications, 122(6), pp.
33-35.
BIBLIOGRAPHY
347
BIEM, A. and CASWELL, N., 2008. A Value Network Model for Strategic Analysis.
Proceedings of the 41st Annual Hawaii International Conference on System Sciences (HICSS
2008). IEEE, pp. 361.
BINDER, M. and EDWARDS, J.S., 2010. Using grounded theory method for theory
building in operations management research: A study on inter-firm relationship governance.
International Journal of Operations & Production Management, 30(3), pp. 232-259.
BLOIS, K., 2006. The boundaries of the firm—a question of interpretation? Industry and
Innovation, 13(2), pp. 135-150.
BLOMQVIST, K. and KIANTO, A., 2007. Knowledge-based view of the firm-theoretical
notions and implications for management. Department of Business Administration and
Technology Business Research Center, Lappeenranta University of Technology, .
BOHÓRQUEZ, L.E. and ESPINOSA, A., 2015. Theoretical approaches to managing
complexity in organizations: A comparative analysis. Estudios Gerenciales, 31(134), pp. 20-29.
BONABEAU, E., 2002. Agent-based modeling: Methods and techniques for simulating
human systems. Proceedings of the National Academy of Sciences, 99(suppl 3), pp. 7280-7287.
BONCHEV, D. and BUCK, G.A., 2005. Quantitative measures of network complexity. In:
D. BONCHEV and D. ROUVRAY, eds, Complexity in Chemistry, Biology, and Ecology. . Springer US,
pp. 191-235.
BOOGERD, F.C., BRUGGEMAN, F.J., HOFMEYR, J.S. and WESTERHOFF, H.V., 2007.
Towards philosophical foundations of Systems Biology: introduction. Systems Biology,
Philosophical foundations, , pp. 3-19.
BOSCHETTI, F., 2008. Mapping the complexity of ecological models. ecological
complexity, 5(1), pp. 37-47.
BOULDING, K.E., 1956. General systems theory-the skeleton of science. Management
science, 2(3), pp. 197-208.
BOWMAN, C. and AMBROSINI, V., 2000. Value creation versus value capture: towards a
coherent definition of value in strategy. British Journal of Management, 11(1), pp. 1-15.
BRANDENBURGER, A.M. and STUART, H.W., 1996. Value-based business strategy.
Journal of Economics and Management Strategy, 5, pp. 5-24.
BRANDES, U., RAAB, J. and WAGNER, D., 2001. Exploratory network visualization:
Simultaneous display of actor status and connections. Journal of Social Structure, 2(4), pp. 1-28.
BIBLIOGRAPHY
348
BRUSTEIN, J., 2014-last update, How an Online Survey Company Joined the $1 Billion
Club. Available: http://www.bloomberg.com/bw/articles/2014-09-24/how-qualtrics-turned-
online-surveys-into-a-1-billion-business.
BUNGE, M.A., 1967. Scientific Research: The search for truth. Springer-Verlag.
CACUCI, D.G., IONESCU-BUJOR, M. and NAVON, I.M., 2005. Sensitivity and uncertainty
analysis, volume II: applications to large-scale systems. CRC Press.
CALVEZ, B. and HUTZLER, G., 2006. Automatic tuning of agent-based models using
genetic algorithms. In: J. SICHMAN and L. ANTUNES, eds, Multi-Agent-Based Simulation VI. .
Springer Berlin Heidelberg, pp. 41-57.
CAMPBELL, A., 1997. Stakeholders: the case in favour. Long range planning, 30(3), pp.
446-449.
CANTWELL, J.A., 2013. Blurred boundaries between firms, and new boundaries within
(large multinational) firms: the impact of decentralized networks for innovation. Seoul Journal of
Economics, 26, pp. 1-32.
CAPRA, F., 1985. The Tao of Physics. Boulder, Colorado: Shambhala.
CARRILLO-HERMOSILLA, J. and UNRUH, G.C., 2006. Technology stability and change: an
integrated evolutionary approach. Journal of Economic Issues, 40(3), pp. 707-742.
CARRILLO-HERMOSILLA, J., 2015. Technological Diffusion and Standardization Patterns:
An Industrial Taxonomy. Journal of Economic Issues, 49(1), pp. 253-263.
CASTELLETTI, A. and SONCINI-SESSA, R., 2007. Bayesian Networks and participatory
modelling in water resource management. Environmental Modelling & Software, 22(8), pp.
1075-1088.
CASTILLO, O., MELIN, P., KACPRZYK, J. and PEDRYCZ, W., 2006. Hybrid Intelligent
Systems: Analysis and Design (Studies in Fuzziness and Soft Computing). Springer-Verlag New
York, Inc.
CAUDILL, M. and BUTLER, C., 1992. Naturally intelligent systems. MIT Press.
CHAMBERLIN, E.H., 1949. The theory of monopolistic competition: A re-orientation of
the theory of value.
CHANDLER, A.D., 1962. Strategy and structure: Chapters in the history of the American
enterprise. Massachusetts Institute of Technology Cambridge, .
CHANDLER, A.D, 1977. The Visible Hand. Cambridge MA, .
BIBLIOGRAPHY
349
CHANDLER, A.D., 1990. Scale and scope: The dynamics of industrial competition.
Cambridge, MA, Harvard Business School, .
CHANG, M. and HARRINGTON, J.E., 2006. Agent-based models of organizations.
Handbook of computational economics, 2, pp. 1273-1337.
CHEN, S.H. and POLLINO, C.A., 2012. Good practice in Bayesian network modelling.
Environmental Modelling & Software, 37, pp. 134-145.
CLAUSET, A., MOORE, C. and NEWMAN, M.E.J., 2008. Hierarchical structure and the
prediction of missing links in networks. Nature, 453(7191), pp. 98-101.
COASE, R.H., 1937. The Nature of the Firm. Economica, 4(16), pp. 386-405.
COCKBURN, D. and KOBTI, Z., 2009. Agent specialization in complex social swarms.
Innovations in Swarm Intelligence. . Springer, pp. 77-89.
COLLINS, J.C. and PORRAS, J.I., 1994. Built to Last: Successful Habits of Visionary
Companies. Harper Business.
COOK, K.S. and EMERSON, R.M., 1987. Social exchange theory. SAGE Publications.
COUGHLAN, P. and COUGHLAN, D., 2002. Action research for operations management.
International journal of operations & production management, 22(2), pp. 220-240.
CRUTCHFIELD, J., 2008. Five questions on complexity: Responses. In: C. GERSHENSON,
ed, Complexity: 5 Questions. Copenhagen, Denmark: Automatic Press /VIP, .
CYERT, R.M. and MARCH, J.G., 1963. A behavioral theory of the firm. Englewood Cliffs,
NJ, 2.
CYERT, R.M. and HEDRICK, C.L., 1972. Theory of the Firm: Past, Present, and Future; An
Interpretation. Journal of Economic Literature, 10(2), pp. 398-412.
DANEKE, G.A., 1997. From metaphor to method: Nonlinear science and practical
management. The International Journal of Organizational Analysis, 5(3), pp. 249-266.
DAVIS, B. and SUMARA, D., 2014. Complexity and Education: Inquiries Into Learning,
Teaching, and Research. Taylor & Francis.
DAY, E., 2002. The Role of Value in Consumer Satisfaction. Journal of Consumer
Satisfaction, Dissatisfaction and Complaining Behavior, 15, pp. 22-32.
DE TONI, A.F., NARDINI, A., NONINO, F. and ZANUTTO, G., 2001. Complexity measures in
manufacturing systems.
BIBLIOGRAPHY
350
DEAN, J.W., OTTENSMEYER, E. and RAMIREZ, R., 1997. An aesthetic perspective on
organizations. In: L. COOPER and S.E. JACKSON, eds, Creating Tomorrow's Organizations: A
Handbook for Future Research in Organizational Behavior. Chichester: Wiley, pp. 419-438.
DEHMER, M., 2011. Information theory of networks. Symmetry, 3, pp. 767-779.
DEMSETZ, H., 1988. Ownership, control, and the firm. Oxford: Blackwell.
DESAI, V.S., CROOK, J.N. and OVERSTREET, G.A., 1996. A comparison of neural networks
and linear scoring models in the credit union environment. European Journal of Operational
Research, 95(1), pp. 24-37.
DESARBO, W.S., JEDIDI, K. and SINHA, I., 2001. Customer value analysis in a
heterogeneous market. Strategic Management Journal, 22(9), pp. 845-857.
DESHMUKH, A.V., TALAVAGE, J.J. and BARASH, M.M., 1998. Complexity in
manufacturing systems, part 1: Analysis of static complexity. IIE Transactions, 30(7), pp. 645-
655.
DIETRICH, M. and KRAFFT, J., 2012. Handbook on the Economics and Theory of the Firm.
Edward Elgar Publishing.
DONALDSON, T. and PRESTON, L.E., 1995. The Stakeholder Theory of the Corporation:
Concepts, Evidence, and Implications. The Academy of Management Review, 20(1), pp. 65-91.
EDITORIAL, 2009. No man is an island. Nature Physics, 5(1),.
EIDELSON, R.J., 1997. Complex adaptive systems in the behavioral and social sciences.
Review of General Psychology, 1(1), pp. 42-71.
ETHIRAJ, S.K. and LEVINTHAL, D., 2004. Modularity and innovation in complex systems.
Management Science, 50(2), pp. 159-173.
FAGIOLO, G., 2007. Clustering in complex directed networks. Physical Review E, 76(2),.
FEO, T.A. and RESENDE, M.G., 1995. Greedy randomized adaptive search procedures.
Journal of Global Optimization, 6(2), pp. 109-133.
FLOUDAS, C.A. and PARDALOS, P.M., 2013. State of the art in global optimization:
computational methods and applications. Springer Science & Business Media.
FORRESTER, J.W., 1987. Lessons from system dynamics modeling. System Dynamics
Review, 3(2), pp. 136-149.
FORRESTER, J.W., 1994. System dynamics, systems thinking, and soft OR. System
Dynamics Review, 10(2‐3), pp. 245-256.
BIBLIOGRAPHY
351
FOSS, N.J., 1998. The theory of the firm: an introduction to themes and contributions.
Institut for Industrikonomi og Virksomhedsstrategi.
FOSS, N.J. and KLEIN, P.G., 2005. The theory of the firm and its critics: a stocktaking and
assessment. Available at SSRN 695484, .
FREEMAN, L.C., 2000. Visualizing social networks. Journal of Social Structure, 1(1),.
FREEMAN, R.E., 1984. Strategic management: A stakeholder approach. 1st Ed edn.
Pitman.
FREEMAN, R.E., WICKS, A.C. and PARMAR, B., 2004. Stakeholder Theory and The
Corporate Objective Revisited. Organization Science, 15(3), pp. 364-369.
FRIESZ, T.L. and BERNSTEIN, D., 2015. Foundations of Network Optimization and Games.
Springer.
GARRISON, R.H., NOREEN, E.W. and BREWER, P.C., 2006. Managerial accounting.
McGraw-Hill/Irwin.
GAVETTI, G., GREVE, H.R., LEVINTHAL, D.A. and OCASIO, W., 2012. The behavioral theory
of the firm: Assessment and prospects. The academy of management annals, 6(1), pp. 1-40.
GELL-MANN, M., 1995. What is complexity? Complexity, 1(1),.
GHARAJEDAGHI, J., 2011. Systems thinking: Managing chaos and complexity: A platform
for designing business architecture. Elsevier.
GHOSHAL, S. and MORAN, P., 1997. Value Creation By Firms. INSEAD.
GIBBONS, M., LIMOGES, C., NOWOTNY, H., SCHWARTZMAN, S., SCOTT, P. and TROW,
M., 1994. The new production of knowledge: The dynamics of sand Research in contemporary
societies. SAGE Publications.
GILBERT, N. and TROITZSCH, K., 2005. Simulation for the social scientist. McGraw-Hill
Education (UK).
GLASER, B.G. and STRAUSS, A.L., 2009. The discovery of grounded theory: Strategies for
qualitative research. Transaction Publishers.
GLOVER, F., 1989. Tabu Search—Part I. ORSA Journal on Computing, 1(3), pp. 190-206.
GNATZY, T., WARTH, J., VON DER GRACHT, H. and DARKOW, I., 2011. Validating an
innovative real-time Delphi approach-A methodological comparison between real-time and
conventional Delphi studies. Technological Forecasting and Social Change, 78(9), pp. 1681-1694.
BIBLIOGRAPHY
352
GODFREY, P.C. and HILL, C.W.L., 1995. The problem of unobservables in strategic
management research. Strategic Management Journal, 16(7), pp. 519-533.
GOERZEN, A. and BEAMISH, P.W., 2007. The Penrose effect:“Excess” expatriates in
multinational enterprises. Management International Review, 47(2), pp. 221-239.
GOFFE, W.L., FERRIER, G.D. and ROGERS, J., 1994. Global optimization of statistical
functions with simulated annealing. Journal of Econometrics, 60(1), pp. 65-99.
GOLDENFELD, N. and KADANOFF, L.P., 1999. Simple lessons from Complexity. Science,
(284), pp. 87-89.
GORZEŃ-MITKA, I. and OKRĘGLICKA, M., 2015. Managing Complexity: A Discussion of
Current Strategies and Approaches. Procedia Economics and Finance, 27, pp. 438-444.
GRANT, R.M., 1996. Toward a knowledge‐based theory of the firm. Strategic
Management Journal, 17(S2), pp. 109-122.
GRIMM, V., BERGER, U., BASTIANSEN, F., ELIASSEN, S., GINOT, V., GISKE, J., GOSS-
CUSTARD, J., GRAND, T., HEINZ, S.K., HUSE, G., HUTH, A., JEPSEN, J.U., JØRGENSEN, C., MOOIJ,
W.M., MÜLLER, B., PE’ER, G., PIOU, C., RAILSBACK, S.F., ROBBINS, A.M., ROBBINS, M.M.,
ROSSMANITH, E., RÜGER, N., STRAND, E., SOUISSI, S., STILLMAN, R.A., VABØ, R., VISSER, U. and
DEANGELIS, D.L., 2006. A standard protocol for describing individual-based and agent-based
models. Ecological Modelling, 198(1–2), pp. 115-126.
GRZEGORCZYK, M. and HUSMEIER, D., 2009. Bayesian networks and their applications in
systems biology. and Posters, 17, pp. 48.
GULATI, R., 1998. Alliances and networks. Strategic Management Journal, 19(4), pp.
293-317.
HAGAN, M., DEMUTH, H., BEALE, M. and DE JESUS, O., 2014. Neural Network Design
(2nd Edition). Martin Hagan.
HAILE, N. and ALTMANN, J., 2013. Estimating the value obtained from using a software
service platform. Economics of Grids, Clouds, Systems, and Services. . Springer, pp. 244-255.
HAKANSSON, H. and SNEHOTA, I., 2006. No business is an island: The network concept
of business strategy. Scandinavian Journal of Management, 22(3), pp. 256-270.
HAKSEVER, C., CHAGANTI, R. and COOK, R.G., 2004. A model of value creation: Strategic
view. Journal of Business Ethics, 49(3), pp. 295-307.
HARRISON, J.S. and WICKS, A.C., 2013. Stakeholder theory, value, and firm performance.
Business ethics quarterly, 23(01), pp. 97-124.
BIBLIOGRAPHY
353
HART, O. and MOORE, J., 1990. Property Rights and the Nature of the Firm. Journal of
political economy, , pp. 1119-1158.
HEILBRONER, R.L., 1983. The Problem of Value in the Constitution of Economic Thought.
Social Research, 50(2), pp. 253-277.
HEISENBERG, W.K., 1958. Physics and Philosophy. The Revolution in Modern Science.
Based on The Gifford Lectures delivered at the University of St. Andrews during the Winter Term
1955-1956.
HELANDER, N., 2004. Value-creating networks: an analysis of the software component
business. University of Oulu.
HELBING, D., 2012. Social Self-organization: Agent-based simulations and experiments to
study emergent social behavior. Springer.
HELPER, S., 2000. Economists and field research:" You can observe a lot just by
watching". American Economic Review, 90(2), pp. 228-232.
HERGERT, M. and MORRIS, D., 1989. Accounting data for value chain analysis. Strategic
Management Journal, 10(2), pp. 175-188.
HERRALA, M., PAKKALA, P. and HAAPASALO, H., 2011. Value-creating networks. A
conceptual model and analysis. Department of Industrial Engineering and Management,
University of Oulu.
HILLMAN, A.J. and KEIM, G.D., 2001. Shareholder value, stakeholder management, and
social issues: what's the bottom line? Strategic Management Journal, 22(2), pp. 125-139.
HINKELMANN, F., MURRUGARRA, D., JARRAH, A.S. and LAUBENBACHER, R., 2010. A
mathematical framework for agent based models of complex biological networks. Bulletin of
Mathematical Biology, .
HITT, M.A., KEATS, B.W. and DEMARIE, S.M., 1998. Navigating in the new competitive
landscape: Building strategic flexibility and competitive advantage in the 21st century. Academy
of Management, 12(4), pp. 22-42.
HITT, M.A., GIMENO, J. and HOSKISSON, R.E., 1998. Current and Future Research
Methods in Strategic Management. Organizational Research Methods, 1(1), pp. 6-44.
HOLLAND, J.H., 1975. Adaptation in natural and artificial systems: an introductory
analysis with applications to biology, control, and artificial intelligence. University of Michigan
Press.
HOLLAND, J.H., 2000. Emergence: From chaos to order. Oxford University Press.
BIBLIOGRAPHY
354
HÖLMSTROM, B., 1979. Moral hazard and observability. The Bell journal of economics, ,
pp. 74-91.
HÖLMSTROM, B. and MILGROM, P., 1994. The firm as an incentive system. The
American Economic Review, , pp. 972-991.
HOLMSTRÖM, B. and ROBERTS, J., 1998. The boundaries of the firm revisited. The
Journal of Economic Perspectives, , pp. 73-94.
HOLWEG, M. and PIL, F.K., 2006. Evolving from value chain to value grid. MIT Sloan
Management Review, 47(4), pp. 72-80.
HOSKISSON, R.E., HITT, M.A., WAN, W.P. and YIU, D., 1999. Theory and research in
strategic management: Swings of a pendulum. Journal of management, 25(3), pp. 417-456.
HOU-SHUN, L., 1956. The Concept of Economic Homeostasis. Financial Analysts Journal,
12(4), pp. 51-53.
HOWARD, N., ROLLAND, C. and QUSAIBATY, A., 2004. Process Complexity: Towards a
Theory of Intentoriented Process Design. 2nd International Conference of Information and
Systems. Egypt: ICIS, .
IANSITI, M. and LEVIEN, R., 2004. The keystone advantage: what the new dynamics of
business ecosystems mean for strategy, innovation, and sustainability. Harvard Business Press.
INDIANA, U., 2005-last update, Pathfinder Network Scaling. Available: Information
Visualization Cyberinfrastructure http://iv.slis.indiana.edu/.
INDIANA, U., NORTHEASTERN, U. and MICHIGAN, U., 2006-last update, Network
Workbench Tool. Available: http://nwb.slis.indiana.edu.
INTERNATIONAL AIR TRANSPORT ASSOCIATION, 2011. Vision 2050. Available in:
www.iata.org, .
ITAMI, H. and ROEHL, T.W., 1991. Mobilizing invisible assets. Harvard University Press.
IZQUIERDO, L.R., ORDAX, J.M.G., SANTOS, J.I. and MARTÍNEZ, R.D.O., 2008. Modelado
de sistemas complejos mediante simulación basada en agentes y mediante dinámica de
sistemas. Empiria. Revista de metodología de ciencias sociales, (16), pp. 85-112.
JACCARD, J. and JACOBY, J., 2010. Theory construction and model-building skills: A
practical guide for social scientists. Guilford Press.
JACKSON, M.O., 2008. Social and economic networks. Princeton University Press
Princeton.
BIBLIOGRAPHY
355
JANG, J.S.R., SUN, C.T. and MIZUTANI, E., 1997. Neuro-fuzzy and Soft Computing: A
Computational Approach to Learning and Machine Intelligence. Prentice Hall.
JARVENPAA, S.L. and IVES, B., 1994. The global network organization of the future:
Information management opportunities and challenges. Journal of Management Information
Systems, 10(4), pp. 25-57.
JÄRVINEN, P., 2005. Action research as an approach in design science. University of
Tampere, Retrieved from http://www.sis.uta.fi/cs/reports/dsarja/D-2005-2.pdf.
JAYNES, E.T., 1991. How Should We Use Entropy In Economics? .
JENSEN, M.C. and MECKLING, W.H., 1992. Specific and general knowledge and
organizational structure. Available at SSRN 6658, .
JENSEN, M.C. and MECKLING, W.H., 2000. The theory of the firm: managerial behavior,
agency costs and ownership structure. Theory of the Firm, Bd, 1, pp. 248-306.
JENSEN, M.C., 2002. Value Maximization, Stakeholder Theory, and the Corporate
Objective Function. Business Ethics Quarterly, 12(2), pp. 235-256.
JETTER, A.J. and KOK, K., 2014. Fuzzy Cognitive Maps for futures studies—A
methodological assessment of concepts and methods. Futures, 61, pp. 45-57.
JOHNSON, R.A., KAST, F.E. and ROSENZWEIG, J.E., 1973. The theory and management of
systems. McGraw-Hill.
JOHNSTON, R. and LAWRENCE, P.R., 1988. Beyond vertical integration -- the rise of the
value-adding partnership. Harvard Business Review, .
JØRGENSEN, S.E. and BENDORICCHIO, G., 2001. Fundamentals of ecological modelling.
Elsevier.
JØRGENSEN, S.E., 2006. An integrated ecosystem theory. Annals of the European
Academy of Science, , pp. 19-33.
JOYCE, K.E., HAYASAKA, S. and LAURIENTI, P.J., 2012. A genetic algorithm for controlling
an agent-based model of the functional human brain. Biomed Sci Instrum, 48, pp. 210-217.
KANEKO, K., 2006. Life: an introduction to complex systems biology. Springer.
KAPLAN, R.S. and NORTON, D.P., 1996. The balanced scorecard: Translating strategy into
action. Harvard Business School Press.
KAPLAN, R.S. and NORTON, D.P., 2004. Strategy maps: Converting intangible assets into
tangible outcomes. Harvard Business Press.
BIBLIOGRAPHY
356
KAPOOR, R. and LEE, J.M., 2013. Coordinating and competing in ecosystems: How
organizational forms shape new technology investments. Strategic Management Journal, 34(3),
pp. 274-296.
KARSTEN, S.G., 1990. Quantum theory and social economics: The holistic approach of
modern physics serves better than Newton's mechanics in approaching reality. American Journal
of Economics and Sociology, , pp. 385-399.
KAST, F.E. and ROSENZWEIG, J.E., 1972. General Systems Theory: Applications for
Organization and Management. The Academy of Management Journal, 15(4, General Systems
Theory), pp. 447-465.
KAUFFMAN, S.A., 1990. The Sciences of Complexity and" Origins of Order". Proceedings
of the Biennial Meeting of the Philosophy of Science Association. . JSTOR, pp. 299-322.
KHULLER, S. and RAGHAVACHARI, B., 2013. Network Optimization. In: P.M. PARDALOS,
D. DU and R.L. GRAHAM, eds, Springer New York, pp. 1989-2026.
KITANO, H., 2002. Systems biology: toward system-level understanding of biological
systems. In: H. KITANO, ed, Foundations of systems biology. Cambridge, MA: MIT Press, pp. 1-36.
KOHONEN, T., 1998. The self-organizing map. Neurocomputing, 21(1), pp. 1-6.
KOHONEN, T., 2013. Essentials of the self-organizing map. Neural Networks, 37(Twenty-
fifth Anniversary Commemorative Issue), pp. 52-65.
KOLACZYK, E.D., 2009. Statistical Analysis of Network Data. Methods and Models.
Springer.
KOLACZYK, E.D. and CSÁRDI, G., 2014. Statistical analysis of network data with R.
Springer.
KOLLER, T., GOEDHART, M. and WESSELS, D., 2010. Valuation: measuring and managing
the value of companies. john Wiley and sons.
KOSKO, B., 1987. Adaptive inference in fuzzy knowledge networks. Proc. 1st Int. Conf.
Neural Networks. pp. 261-268.
KOTHANDARAMAN, P. and WILSON, D.T., 2001. The future of competition: value-
creating networks. Industrial marketing management, 30(4), pp. 379-389.
KRIESEL, D., 2007. A brief introduction to neural networks. Retrieved from:
http://www.dkriesel.com.
KRYCHA, K.A. and WAGNER, U., 1999. Applications of artificial neural networks in
management science: a survey. Journal of Retailing and Consumer Services, 6(4), pp. 185-203.
BIBLIOGRAPHY
357
KUDRYAVTSEV, L.D., 2001-last update, Homogeneous function. Available:
http://www.encyclopediaofmath.org/index.php?title=Homogeneous_function&oldid=11366.
LADYMAN, J., LAMBERT, J. and WIESNER, K., 2013. What is a complex system? European
Journal for Philosophy of Science, 3(1), pp. 33-67.
LAMBERTINI, L., 2013. John von Neumann between Physics and Economics: A
Methodological Note. Review of Economic Analysis, 5, pp. 177–189.
LEE, K.C., LEE, H., LEE, N. and LIM, J., 2013. An agent-based fuzzy cognitive map
approach to the strategic marketing planning for industrial firms. Industrial Marketing
Management, 42(4), pp. 552-563.
LEONDES, C.T., 1998. Optimization Techniques. Elsevier Science.
LEPAK, D.P., SMITH, K.G. and TAYLOR, M.S., 2007. Introduction to special topic forum:
Value creation and value capture. A multilevel perspective. The Academy of Management
Review, 32(1), pp. 180-194.
LEVINTHAL, D., 1988. A survey of agency models of organizations. Journal of Economic
Behavior & Organization, 9(2), pp. 153-185.
LEVINTHAL, D., 2007. Technology: the role of network structures. Strategic
Entrepreneurship Journal, 1(3‐4), pp. 189-190.
LI, E.Y., 1994. Artificial neural networks and their business applications. Information &
Management, 27(5), pp. 303-3013.
LIEBERMAN, M.B. and BALASUBRAMANIAN, N., 2007. Measuring value creation and its
distribution among stakeholders of the firm. Available at SSRN 2382099, .
LIMA, F.W., HADZIBEGANOVIC, T. and STAUFFER, D., 2009. Evolution of ethnocentrism
on undirected and directed Barabási–Albert networks. Physica A: Statistical Mechanics and its
Applications, 388(24), pp. 4999-5004.
LINSTONE, H.A. and TUROFF, M., 2002. The Delphi Method. Techniques and
applications, 53.
LIPPMANN, R.P., 1987. An introduction to computing with neural nets. ASSP Magazine,
IEEE, 4(2), pp. 4-22.
LUENBERGER, D.G., 1979. Introduction to Dynamic Systems: Theory, Models, and
Applications. Wiley.
M’CHIRGUI, Z., 2012. Small-world or Scale-Free Phenomena in Internet: What
Implications for the Next-generation Networks? Review of European Studies, 4(1), pp. p85.
BIBLIOGRAPHY
358
MACAL, C.M. and NORTH, M.J., 2009. Agent-based modeling and simulation.
Proceedings of the 2009 Winter Simulation Conference (WSC). IEEE, pp. 86-98.
MAHONEY, J.T., 2005. Economic foundations of strategy. SAGE Publications.
MANGIAMELI, P., CHEN, S.K. and WEST, D., 1996. A comparison of SOM neural network
and hierarchical clustering methods. European Journal of Operational Research, 93(2), pp. 402-
417.
MARSHALL, A., 1919. Industry and Trade, a Study of Industrial Technique and Business
Organization. Macmillan.
MARSHALL, A. and GUILLEBAUD, C.W., 1961. Principles of Economics. 9th (variorum) ed.
Macmillan.
MARTIN, R., 2012. Opening Up The Boundaries of the Firm. Harvard Business Review, .
MASTEN, S.E., 1984. The organization of production: Evidence from the aerospace
industry. Journal of Law and Economics, , pp. 403-417.
MAULA, M., 2006. Organizations as learning systems:'living composition'as an enabling
infrastructure. Emerald Group Publishing.
MCGARVEY, B., HANNON, B. and HANNON, B.M., 2004. Dynamic modeling for business
management: An introduction. Springer Science & Business Media.
MCKAY, R.S., 2008. Nonlinearity in complexity science. Nonlinearity, (21), pp. T273–
T281.
MCMILLAN, E., 2008. Complexity, management and the dynamics of change: Challenges
for practice. Routledge.
MELLA, P., 2009. The Holonic Revolution. Holons, Holarchies and Holonic Networks. The
Ghost in the Production Machine. Pavia University Press.
MENDOZA, G.A. and PRABHU, R., 2006. Participatory modeling and analysis for
sustainable forest management: Overview of soft system dynamics models and applications.
Forest Policy and Economics, 9(2), pp. 179-196.
MEYER, A.D., GABA, V. and COLWELL, K.A., 2005. Organizing far from equilibrium:
Nonlinear change in organizational fields. Organization Science, 16(5), pp. 456-473.
MEYER, R.R., 1985. Network Optimization. In: K. SCHITTKOWSKI, ed, Springer Berlin
Heidelberg, pp. 125-139.
MITCHELL, M., 1998. An introduction to genetic algorithms. Bradford Books.
BIBLIOGRAPHY
359
MOODY, J., MCFARLAND, D. and BENDER‐DEMOLL, S., 2005. Dynamic network
visualization. American Journal of Sociology, 110(4), pp. 1206-1241.
MORGENBESSER, S., 1966. Is It a Science? Social Research, 33(2), pp. 255.
MOSS, S., 1984. The History of the Theory of the Firm from Marshall to Robinson and
Chamberlin: the Source of Positivism in Economics. Economica, , pp. 307-318.
MOTTER, A.E. and TOROCZKAI, Z., 2007. Introduction: Optimization in networks. Chaos,
17(026101),.
MOWSHOWITZ, A. and DEHMER, M., 2010. A symmetry index for graphs. Journal of
Mathematical Biophysics, 30, pp. 533-546.
NAGARAJAN, R., SCUTARI, M. and LÈBRE, S., 2014. Bayesian Networks in R: with
Applications in Systems Biology. Springer New York.
NAMBISAN, S., AGARWAL, R. and TANNIRU, M., 1999. Organizational mechanisms for
enhancing user innovation in information technology. MIS quarterly, , pp. 365-395.
NELSON, R. and WINTER, S.G., 1982. An Evolutionary Theory of Economic Change.
Cambridge, Mass.: Harvard University Press.
NELSON, R.R. and SAMPAT, B.N., 2001. Making sense of institutions as a factor shaping
economic performance. Revista de Economía Institucional, 3(5), pp. 17-51.
NIELSEN, T.D. and JENSEN, F.V., 2009. Bayesian networks and decision graphs. Springer
Science & Business Media.
NOLAN, P., 2007. Capitalism and freedom: the contradictory character of globalisation.
Anthem Press.
NORMANN, R. and RAMIREZ, R., 1993. From value chain to value constellation:
Designing interactive strategy. Harvard Business Review, 71(4),.
NORTH, D.C., 1990. Institutions, Institutional Change and Economic Performance.
Cambridge University Press.
OBERHEIM, E. and HOYNINGEN-HUENE, P., 2013. The Incommensurability of Scientific
Theories. The Stanford Encyclopedia of Philosophy, (Spring 2013),.
OECD, 2008. Report on Applications of Complexity Science for Public Policy: New Tools
for Finding Unanticipated Consequences and Unrealized Opportunities. OECD Global Science
Forum. . OECD, .
BIBLIOGRAPHY
360
OREMLAND, M. and LAUBENBACHER, R., 2014. Optimization of Agent-Based Models:
Scaling Methods and Heuristic Algorithms. Journal of Artificial Societies and Social Simulation,
17(2), pp. 6.
PABLO-MARTÍ, F., MUÑOZ-YEBRA, C. and SANTOS, J.L., 2014. An agent-based model of
firm location with different regional policies, Social Simulation Conference 2014.
PABLO-MARTÍ, F., SANTOS, J.L. and KASZOWSKA, J., 2015. An agent-based model of
population dynamics for the European regions. Emergence: Complexity and Organization, 17(2),
pp. C1.
PALIWAL, M. and KUMAR, U.A., 2009. Neural networks and statistical techniques: A
review of applications. Expert Systems with Applications, 36(1), pp. 2-17.
PANNELL, D.J., 1997. Sensitivity analysis of normative economic models: theoretical
framework and practical strategies. Agricultural economics, 16(2), pp. 139-152.
PANWAI, S. and DIA, H., 2005. A reactive agent-based neural network car following
model. Proceedings. 2005 IEEE Intelligent Transportation Systems, 2005. . IEEE, pp. 375-380.
PAPAGEORGIOU, E., 2012. Learning algorithms for fuzzy cognitive maps—a review
study. Systems, Man, and Cybernetics, Part C: Applications and Reviews, IEEE Transactions on,
42(2), pp. 150-163.
PAPAGEORGIOU, E. and SALMERON, J.L., 2013. A review of fuzzy cognitive maps
research during the last decade. Fuzzy Systems, IEEE Transactions on, 21(1), pp. 66-79.
PAPANDREOU, A.G., 2000. Some basic problems in the theory of the firm. The Theory of
the Firm: Critical Perspectives on Business and Management, 1, pp. 14.
PARADICE, D., 2009. Emerging Systems Approaches in Information Technologies:
Concepts, Theories, and Applications: Concepts, Theories, and Applications. IGI Global.
PAROLINI, C., 1996. The Value Nets: A methodology for the analysis of value-creating
systems. Strategic Management Conference. . SMS, .
PARTRIDGE, M.J. and PERREN, L.J., 1994. Cost analysis of the value chain: Another role
for strategic management accounting. Management Accounting, 72(7), pp. 22.
PEARL, J., 2014. Probabilistic reasoning in intelligent systems: networks of plausible
inference. Morgan Kaufmann.
PENROSE, E., 1955. Limits to the Growth and Size of Firms. The American Economic
Review, , pp. 531-543.
BIBLIOGRAPHY
361
PENROSE, E.T., 1959. The Theory of the Growth of the Firm. New York: John Wiley &
Sons.
PEPPARD, J. and RYLANDER, A., 2006. From value chain to value network:: Insights for
mobile operators. European Management Journal, 24(2), pp. 128-141.
PIERCE, L., 2008. Big losses in ecosystem niches: How core firm decisions drive
complementary product shakeouts. Available at SSRN 1157054, .
PIGOU, A.C., 1932. The economics of welfare, 1920. McMillan&Co., London, .
PORTER, M., 1985. Competitive advantage: Creating and sustaining superior
performance. New York: Free Press.
PORTER, M.E. and KRAMER, M.R., 2011. Creating shared value. Harvard business review,
89(1/2), pp. 62-77.
PRATT, J.W. and ZECKHAUSER, R., 1985. Principals and agents: the structure of business.
Harvard Business School Press.
PREVOLNIK, M., ŠKORJANC, D., ČANDEK-POTOKAR, M. and NOVIČ, M., 2011. Application
of artificial neural networks in meat production and technology. Computer and Information
Science, 11, pp. 223-240.
PRIEM, R.L., 2007. A consumer perspective on value creation. Academy of Management
Review, 32(1), pp. 219-235.
QUADDUS, M.A. and KHAN, M.S., 1999. Business applications of artificial neural
networks: an updated review and analysis. Neural Information Processing, 1999. Proceedings.
6th International Conference on. Perth, WA: ICONIP'99, pp. 824 vol.2.
RAMÍREZ, R., 1999. Value Co-Production: Intellectual Origins and Implications for
Practice and Research. Strategic Management Journal, 20(1), pp. 49-65.
RAO RAGHURAJ, K. and LAKSHMINARAYANAN, S., 2006. Alternate complexity measures
and stability analysis of process and biological networks. Proceedings of the 11th APCChE
Congress. Kuala Lumpur: APCChE, .
RASHEVSKY, N., 1955. Life, information theory, and topology. The Bulletin of
mathematical biophysics, 17(3), pp. 229-235.
RAUCH, W., 1979. The decision delphi. Technological Forecasting and Social Change,
15(3), pp. 159-169.
RENCHER, A.C. and CHRISTENSEN, W.F., 2012. Methods of Multivariate Analysis. Wiley.
BIBLIOGRAPHY
362
RICHARDSON, G.B., 1972. The organisation of industry. The economic journal, , pp. 883-
896.
RICHARDSON, K. and CILLIERS., P., 2001. Special Editors' Introduction: What Is
Complexity Science? A View from Different Directions. Emergence, 3(1), pp. 5-23.
RIND, D., 1999. Complexity and the climate. Science, (284), pp. 105-107.
ROBINSON, A., 1934. The problem of management and the size of firms. The Economic
Journal, , pp. 242-257.
RODIN, V., QUERREC, G., BALLET, P., BATAILLE, F., DESMEULLES, G., ABGRALL, J. and
TISSEAU, J., 2009. Multi-Agents System to Model Cell Signalling by Using Fuzzy Cognitive Maps.
Application to Computer Simulation of Multiple Myeloma. Ninth IEEE International Conference
on Bioinformatics and BioEngineering. . IEEE, pp. 236-241.
ROOKE, J., MOLLOY, E., SINCLAIR, M., KOSKELA, L., SIRIWARDENA, M., KAGIOGLOU, M.
and SIEMIENIUCH, C., 2008. Models and metaphors: complexity theory and through-life
management in the built environment. Architectural Engineering and Design Management, 4(1),
pp. 47-57.
ROSS, S.A., 1973. The economic theory of agency: The principal's problem. The American
Economic Review, , pp. 134-139.
ROUSE, W.B., 2007. Complex engineered, organizational and natural systems. Systems
Engineering, 10(3), pp. 260-271.
SALTELLI, A., 2002. Sensitivity analysis for importance assessment. Risk Analysis, 22(3),
pp. 579-590.
SANCHEZ, R. and HEENE, A., 2004. The new strategic management. New York, .
SANDERS, T.I., 2003-last update, What is complexity?. Available:
http://www.complexsys.org.
SCHMIDT, R., LYYTINEN, K. and MARK KEIL, P.C., 2001. Identifying software project risks:
An international Delphi study. Journal of Management Information Systems, 17(4), pp. 5-36.
SCHOCKEN, S. and ARIAV, G., 1994. Neural networks for decision support: Problems and
opportunities. Decision Support Systems, 11(5), pp. 393-414.
SCOTT, W.R. and DAVIS, G.F., 2007. Organizations and organizing: Rational, natural, and
open system perspectives. Prentice Hall.
SCOTT, W.G., 1961. Organization theory: an overview and an appraisal. Journal of the
Academy of Management, , pp. 7-26.
BIBLIOGRAPHY
363
SERRANO-CINCA, C., 1996. Self organizing neural networks for financial diagnosis.
Decision Support Systems, 17(3), pp. 227-238.
SETIONO, R., THONG, J.Y. and YAP, C., 1998. Symbolic rule extraction from neural
networks: An application to identifying organizations adopting IT. Information & management,
34(2), pp. 91-101.
SEXTON, R.S., DORSEY, R.E. and JOHNSON, J.D., 1999. Optimization of neural networks:
A comparative analysis of the genetic algorithm and simulated annealing. European Journal of
Operational Research, 114(3), pp. 589-601.
SHARDA, R. and WANG, J., 1996. Neural networks and operations
research/management science. European Journal of Operational Research, 93(2), pp. 227-229.
SIMON, H.A., 1953. Administrative behavior: a study of decision-making processes in
administrative organization. Macmillan.
SIMON, H.A., 1962a. The architecture of complexity. Proceedings of the American
Philosophical Society, Vol. 106, No. 6, pp. 467-482.
SIMON, H.A., 1962b. New Developments in the Theory of the Firm. The American
Economic Review, 52(2, Papers and Proceedings of the Seventy-Fourth Annual Meeting of the
American Economic Association), pp. 1-15.
SIMON, H.A., 1976. How complex are complex systems? PSA: Proceedings of the Biennial
Meeting of the Philosophy of Science Association, 1976(, Volume Two: Symposia and Invited
Papers), pp. 507-522.
SIMON, H.A., 1982. Models of Bounded Rationality: Empirically grounded economic
reason. MIT Press.
SIMON, H.A., 1996. The sciences of the artificial. MIT press.
SIMON, H.A., 1999. Can there be a science of complex systems? In: Y. BAR-YAM, ed,
Unifying themes in complex systems: Proceedings from the International Conference on Complex
Systems 1997. Cambridge, MA: Perseus Press, pp. 4-14.
SKULMOSKI, G.J., HARTMAN, F.T. and KRAHN, J., 2007. The Delphi Method for Graduate
Research. Journal of Information Technology for Teacher Education [H.W. Wilson - EDUC], 6, pp.
1.
SLOCUM, N., 2005. Delphi: Participatory Methods Toolkit. Retrieved from
http://archive.unu.edu/hq/library/Collection/PDF_files/CRIS/PMT.pdf.
BIBLIOGRAPHY
364
SMITH, K.A. and GUPTA, J.N.D., 2000. Neural networks in business: techniques and
applications for the operations researcher. Computers & Operations Research, 27(11-12), pp.
1023-1044.
SMITH, K.A. and GUPTA, J.N., 2002. Neural networks in business: techniques and
applications. IGI Global.
SMITH, P.A. and MITLETON-KELLY, E., 2011. A complexity theory approach to
sustainability: A longitudinal study in two London NHS hospitals. The Learning Organization,
18(1), pp. 45-53.
SONG, H., MIAO, C., ROEL, W., SHEN, Z. and CATTHOOR, F., 2010. Implementation of
fuzzy cognitive maps based on fuzzy neural network and application in prediction of time series.
Fuzzy Systems, IEEE Transactions on, 18(2), pp. 233-250.
STABELL, C.B. and FJELDSTAD, ØD., 1998. Configuring value for competitive advantage:
on chains, shops, and networks. Strategic Management Journal, 19(5), pp. 413-437.
STACE, W.T. and GOLDSTEIN, J.A., 2006. Novelty, indeterminism, and emergence.
Emergence: Complexity and Organization, 8(2), pp. 77.
STACEY, R.D., GRIFFIN, D. and SHAW, P., 2000. Complexity and management: fad or
radical challenge to systems thinking? Psychology Press.
STACEY, R. and GRIFFIN, D., 2008. Complexity and the experience of values, conflict and
compromise in organizations. Routledge.
STERMAN, J.D., 2000. Business dynamics: systems thinking and modeling for a complex
world. Irwin/McGraw-Hill Boston.
STEYVERS, M., TENENBAUM, J.B., WAGENMAKERS, E. and BLUM, B., 2003. Inferring
causal networks from observations and interventions. Cognitive science, 27(3), pp. 453-489.
STROGATZ, S.H., 2001. Exploring complex networks. Nature, 410(6825), pp. 268-276.
STULA, M., STIPANICEV, D. and BODROZIC, L., 2010. Intelligent modeling with agent‐
based fuzzy cognitive map. International Journal of Intelligent Systems, 25(10), pp. 981-1004.
STYLIOS, C.D. and GROUMPOS, P.P., 1999. Mathematical Formulation of Fuzzy Cognitive
Maps. Proceedings of the 7th Mediterranean Conference on Control and Automation. Haifa,
Israel: MED99, .
SUNDARAM, A.K. and INKPEN, A.C., 2004. The corporate objective revisited.
Organization science, 15(3), pp. 350-363.
BIBLIOGRAPHY
365
TAN, D. and MAHONEY, J.T., 2005. Examining the Penrose effect in an international
business context: The dynamics of Japanese firm growth in US industries. Managerial and
Decision Economics, 26(2), pp. 113-127.
TARRIDE, M.I., 2013. The complexity of measuring complexity. Kybernetes, 42(2), pp.
174-184.
THORELLI, H.B., 1986. Networks: Between markets and hierarchies. Strategic
Management Journal, 7(1), pp. 37-51.
TOWN, J.S. and KYRILLIDOU, M., 2013. Developing a values scorecard. Performance
Measurement and Metrics, 14(1), pp. 7-16.
TWOMMEY, P. and CADMAN, R., 2002. Agent-Based Modelling of Customer Behaviour
in the Telecoms and Media Markets. info, 4, 56-63.
VAN AKEN, J.E., 2004. Management research based on the paradigm of the design
Sciences: The quest for field-tested and grounded technological rules. Journal of Management
Studies, 41(2), pp. 219-246.
VAN HAM, F. and VAN WIJK, J.J., 2004. Interactive Visualization of Small World Graphs.
IEEE Symposium on Information Visualization. . IEEE, pp. 199-206.
VELLIDO, A., LISBOA, P.J.G. and VAUGHAN, J., 1999. Neural networks in business: a
survey of applications (1992–1998). Expert Systems with Applications, 17(1), pp. 51-70.
VOINOV, A. and BOUSQUET, F., 2010. Modelling with stakeholders. Environmental
Modelling & Software, 25(11), pp. 1268-1281.
VON BERTALANFFY, L., 1950. The Theory of Open Systems in Physics and Biology.
Science, No. 2872(New Series, Vol. 111), pp. 29.
VON DER GRACHT, H., 2012. Consensus measurement in Delphi studies: review and
implications for future quality assurance. Technological Forecasting and Social Change, 79(8),
pp. 1525-1536.
VON NEUMANN, J. and MORGENSTERN, O., 1947. Theory of Games and Economic
Behavior. Princeton University Press.
WARNER, B. and MISRA, M., 1996. Understanding neural networks as statistical tools.
The american statistician, 50(4), pp. 284-293.
WASSERMAN, S. and FAUST, K., 1994. Social network analysis: Methods and
applications. Cambridge university press.
BIBLIOGRAPHY
366
WATTS, D.J. and STROGATZ, S.H., 1998. Collective dynamics of 'small-world' networks.
Nature, 393(6684), pp. 440-2.
WATTS, D.J., 2004. The" new" science of networks. Annual review of sociology, , pp. 243-
270.
WEAVER, W., 1991. Science and complexity. Facets of Systems Science. . Springer, pp.
449-456.
WELTER, F., SMALLBONE, D. and VAN GILS, A., 2012. Entrepreneurial Processes in a
Changing Economy: Frontiers in European Entrepreneurship Research. Edward Elgar.
WENG, G., BHALLA, U.S. and IYENGAR, R., 1999. Complexity in biological signaling
systems. Science, (284), pp. 92-96.
WERNERFELT, B., 1984. A resource-based view of the firm. Strategic Management
Journal, 5(2), pp. 171-180.
WILLIAMSON, O.E., 1969. Managerial discretion, organization form, and the multi-
division hypothesis. University of Pennsylvania, Department of Economics.
WILLIAMSON, O.E., 1975. Markets and hierarchies: antitrust analysis and implications.
New York: The Free Press.
WILLIAMSON, O.E., 1985. The Economic Institutions of Capitalism: Firms, markets,
relational Contracting. Free Press.
WILLIAMSON, O.E., 1996. The Mechanisms of Governance. Oxford University Press, USA.
WOODALL, T., 2003. Conceptualising'value for the customer': an attributional, structural
and dispositional analysis. Academy of marketing science review, 2003, pp. 1.
YOUNG, S.M., 1999. Field research methods in management accounting. Accounting
horizons, 13(1), pp. 76-84.
ZARANDI, M.H., HADAVANDI, E. and TURKSEN, I.B., 2012. A hybrid fuzzy intelligent
agent‐based system for stock price prediction. International Journal of Intelligent Systems,
27(11), pp. 947-969.
VITA
367
VITA
Mr. Navarro Meneses is a professional with over 18 years’ experience in
business consulting, having served with “Big Four” consultancy
companies, as well as in his own company. His specialties include large
strategic change management and innovation programs in industries
ranging from airlines and hospitality, to health services and government. Along his career, he
has served as CEO for a regional airline and an aircraft MRO company. He holds a BSc in
Economics from Universidad Autónoma de Madrid, and an Executive MBA from Instituto de
Empresa (Madrid). He is also a licensed airplane commercial pilot and a certified project
manager.
