How Does It Benefit Us to View Business Ecosystems as Complex Adaptive Systems?

The more complex our needs become, the more complex our society becomes. An interconnected world is emerging with interdependent phenomena whose effects are difficult to keep track of. In order to satisfy the complex needs of economic participants, the economy reacts to these environmental conditions by forming ecosystems in which companies form multi-layered structures with other companies and customers, interact with each other and go through learning processes individually as well as collectively. One example of such a structure is the ecosystem around the SAP business platform.

With its business platform, SAP is dependent on various other players in a variety of ways – it is exemplarily dependent on other companies, often specialized consultancies, using the platform provided by SAP to develop services for third-party companies, such as SMEs.

Even if we consider the ecosystem from the SAP example only at a certain point in time, it has a high degree of diversity: Each company has its own goals and capabilities and participates in the value creation in different ways, for example by providing different products/product components or contributing to the value creation process of other companies. If we additionally wanted to trace the emergence of the ecosystem or predict its future evolution, the observable elements and interactions would become even more complex and diverse. This diversity is also reflected in the large number of different ecosystem definitions and publications on the topic, with Bogers (2019) stating that since 1992 there have been more than 300 articles published in high-profile publications on ecosystems, a considerable number of which are based on their own definitions.

While it is good that so much attention is being paid to ecosystems as a phenomenon, the large number of definitions is a hindrance both to research and to the application of the concept in practice. Depending on which definition we use, we focus on certain aspects of an ecosystem and leave others out. However, as long as we cannot say with certainty which of these aspects are really relevant for our context, we run the risk of neglecting relevant findings and not obtaining trustworthy results.

It is also relevant for practice that ecosystems have so far been described almost exclusively qualitatively and hardly quantitatively: For example, there are articles dedicated to the question of what the reasons for building an ecosystem can be and what the process of creating an ecosystem looks like: What is too often left open at the moment, for example, are metrics that help a company assess which is more worthwhile: providing an offering on its own or building an ecosystem. Or how many partners and customers are needed at the beginning so that participation in the ecosystem is attractive for all parties involved.

Our long-term goal at CC Ecosystems is to assess ecosystem processes and their impacts in a valid way that can be quantified and measured. To do this, however, we first need a suitable analytical framework: We need to delimit our object of study in such a way that its complexity is reduced to a manageable level, but at the same time no relevant features of an ecosystem are excluded from study. Basically, in CC Ecosystem we follow the structural approach of Adner (2017) to describe and model mechanisms of an ecosystem. However, in order to analyse processes within the ecosystem and to expand our understanding accordingly, it is additionally necessary to set up an analytical framework for quantifying the processes and their effects of an ecosystem, which serves the purpose of generating measurability as well as explainability of the phenomena of an ecosystem.

The natural sciences use the theory of complex adaptive systems (CAS) to study systems that consist of interconnected elements and have the ability to adapt and learn. We now want to make use of this approach, so that we lay a foundation for the quantification of units in an ecosystem. Furthermore, we create a detailed framework of investigation with regard to the accumulation of values among individual actors by describing the influence of the interaction between them.

In order to find out whether an ecosystem can really be considered to be a CAS, we provide a comparison of the definitions and characteristics of an ecosystem and a CAS. For this purpose, we first define the term ‘system’, followed by the description of an ecosystem on the basis of relevant literature, in order to then go into the description of a CAS. This finally enables the comparison between CAS and ecosystem. Finally, we address whether and to what extent the value in an ecosystem can be described by considering it as a CAS.

What Is a System?

Depending on the object of investigation, each discipline defines the term system differently, which is why there is no uniform definition of a system. Since with the theory of complex, adaptive systems we want to transfer a theory to economics that is used in various natural sciences, we need a definition of a system that is as congruent as possible and can be applied to all sciences. For this purpose, we use the following definition of a system according to Aslaksen (2008) for our investigation:

“A system consists of three related sets:

  • a set of elements;
  • a set of internal interactions between the elements; and
  • a set of external interactions between the elements and the rest of the world.

[…] The external interactions are required, on the one hand, for the system to provide its service; on the other hand, for the system to maintain its operational state.” (S. 76)

The term system can be applied both to systems that can be found physically in the world and to ideas: “[…] we recognize that there is nothing in the system concept that restricts it to physical systems; it applies equally well to concepts or ideas, in which case the elements are simpler (or better known) concepts and the interactions are formal relationships between these elements (i.e., inherent in their definitions) (Aslaksen 2008: p.12).”

The elements of a system can be signals, but also components, variables, objects, individuals and the like, and are described by properties such as their structure or their functions. In the context of business ecosystems, we mainly treat companies as elements. The internal interactions consist of relationships between companies, the external interactions form the interactions between companies and their environment outside the ecosystem. The relationship between the elements and the element properties jointly determine system properties such as complexity, determinacy, non-linearity/linearity and many others (cf. Frey, Bossert 2008: p.6; an explanation of these properties follows below).

According to this definition, it is only possible to speak of a system if there is also an environment from which the system delimits itself. System boundaries result from three factors:

  • Due to the demarcation between element and sub/system
  • By the objective and the object of the investigation
  • Through the perspective of the relationship with other systems, where an interface is formed with them

To illustrate the setting of the system boundary, the object of study we are interested in could be the realization of an economic value proposition between several organizations. In this example, our goal would be to identify which actors we need to realize the value proposition. Our system would consist of the system elements “potential businesses” that can provide the value proposition through their mutual relationships. Accordingly, any elements that do not contribute to the creation of the value proposition are not elements of the system; however, they can potentially influence the system at interfaces with the environment. Within the system, there are subsystems consisting of, for example, companies that have a certain capability. A possible external interaction is, for example, competition with another system.

A system can be characterized as either isolated, closed or open. The characterization depends on whether the system exchanges substances and/or energy with the environment. In an economic context, substances are for example services or products, an example of energy exchange is the exchange of information:

  • Closed or isolated systems: There is neither substance nor energy exchange with the environment.
  • Closed systems: there is only energy exchange with the environment
  • Open systems: there is an exchange of energy and substances with the environment

For the sake of simplicity, an open system can be defined independently of its scientific affiliation as a system that has a non-zero transfer balance at the interfaces to its environment. If we consider this definition in an economic context, not only various services of companies, but also complete companies can migrate from one system to another in the form of immigration and emigration.

Figure 1               Product chain as an open system with the elements further processing, product development, supplier and prefabrication, as well as the internal (blue bidirectional arrows) and external (grey bidirectional arrows) interactions.

Finally, the term model should be defined to prevent future misunderstandings before going into the system properties characterizing a CAS.

Within systems analysis, Imboden and Koch (2003: p. 9) define models as concepts for a simplified representation of a complex system. A model represents the important properties of a system while not taking into account its secondary properties. Mainly, models can be divided into two categories, namely physical and abstract or conceptual models. Physical models represent a copy of objects, for example, a model of a car. In contrast, conceptual models are abstract representations of systems or subsystems that can be used to simulate the object being modeled. Abstract models are used in many sciences, such as mathematical models in the form of differential equations, game-theoretic and stochastic models, but also dynamic system models or economic models, which include both qualitative and non-/stochastic forms (cf. Ljung 1998). For us, this means that different models may embody the same system depending on the purpose of the study. Thus, top-down and bottom-up models are two different approaches that can be used to understand and represent relationships between elements of the same system.

Research methods in CC Ecosystems are based on the research approach design science research (DSR). “[…] [D]esign scientists produce and apply knowledge of tasks or situations in order to create effective artifacts. […] Design science products are of four types, constructs, models, methods, and implementations.” (March, Smith: 1995, p. 253). From the perspective of DSR, March and Smith (1995) define a model as “a set of propositions or statements expressing relationships among constructs. In design activities, models represent situations as problem and solution statements. […] the concern of models is utility, not truth […] A semantic data model, for example, is valuable insofar as it is useful for designing an information system. Certain inaccuracies and abstractions are inconsequential for those purposes. […] Although silent or inaccurate on the details, a model may need to capture the structure of reality in order to be a useful representation” (p. 257). Similarly, according to Starfield and Bleloch (1986: p.186 ff., 1987: p. 474), mathematical models and simulations do not encompass all the details of reality, but still provide a useful tool for tracing concepts and processes. “On the other hand, unless the inaccuracies and abstractions inherent in models are understood, their use can lead to inappropriate actions” (March, Smith, 1995: p. 257). While in the natural sciences, for example, correlation and regression analyses from statistical mathematics are used to calculate which model represents reality as accurately as possible, such methods and models are only used in design science if they have a practical use.

Diverse, More Diverse, Ecosystem Definition?

Interestingly, the definition of a business ecosystem is no more fixed than that of a system. This is not necessarily surprising, as researchers from different disciplines adapt the term to their own object of study, just as they do with systems. For example, Adner (2007) defines an ecosystem as “the alignment structure of the multilateral set of partners that need to interact in order for a focal value proposition to materialize (p. 40)”. The definition component ‘alignment structure’ is further described by Adner (2007) as: “Members of an ecosystem have defined positions and activity flows among them. Alignment is the extent to which there is mutual agreement among the members regarding these positions and flows”(p. 42). For Adner, an actor is a single entity, while several actors together form an ecosystem under the condition of mutual agreement regarding their activities and positions. In contrast, for Jacobides (2018), ecosystems are “groups of firms that must deal with either unique or supermodular complementarities that are nongeneric, requiring the creation of a specific structure of relationships and alignment to create value”.

Given the wide range of definitions, there is a danger of using a specialized definition of an ecosystem, which could lead to other manifestations being neglected. This neglect could later have consequences for future research questions. For example, if we only consider platform ecosystems, we might disregard reciprocal relationships between actors that do not involve a platform, but which may well affect a company’s profit.

One method to minimize this risk and to deal with the abundance of ecosystem definitions is offered by Bertalanffy’s work General System Theory (1968). In his research on systems, Bertalanffy encountered a similar problem: Since there was a large amount of definitions in this area as well, he had to find a way to combine these definitions in a meaningful way in order to be able to formulate a general system theory. To this end, he chose a reduction approach: “Instead of studying first one system, then a second, then a third, and so on, [this approach] goes to the other extreme, considers the set of all conceivable systems and then reduces the set to a more reasonable size. This is the method I have recently followed” (p. 95). Following Bertalanffy, we use this method of reducing the set of ecosystem definitions so that we can classify them into classes with common characteristics. To do so, we draw on the literature review of Rong and Shi (2015), who have already applied this approach to a set of ecosystem definitions by reducing the definitions to their most important features and grouping together author classes that focus on the same features. This has resulted in a total of five author classes. We added the definitions of Adner and Jacobides to two of them.

Author ClassKeywordsDescription
1993, 1996, 2006
Concept cosystem; life cycleMoore describes the co-evolution of firms in the ecosystem as the interplay between competitive and cooperative strategies among different levels of organizations, including core business partners, extended businesses, industry associations, and government agencies, which is an extension of Porter’s model of five driving forces (Porter 1979), and presents the four life-cycle phases of an ecosystem: birth, expansion, authority, and renewal.
Iansiti & Levien
2002, 2004a, 2004b
Iansiti & Richards 2006
Role assignment; Role transformationIansiti and his colleagues extend the ecosystem concept to include different role types, their strategies and functions, and the state of an ecosystem. With the help of the role distributions – keystone player, niche player, dominator, hub landlord – the complexity of the interaction between the actors within the ecosystem can be reduced. They further assume that roles can change over time. How such role transformations occur is not mentioned. By testing the ecosystem for productivity, robustness and niche creation, they give a possible approach to measuring the state.
Adner 2006, 2012
Adner & Kapoor 2010
Kapoor & Lee 2013 Jacobides 2018
Complementarities Trade-off ModularityAdner, Kapoor and Jacobides particularly emphasize the role of complementarities in the form of complementary products, activities, and assets. Complementarities enable better use and marketing of the products of the focal firms and thus create added value. Ecosystems only make sense if there is a significant need for coordination between diverse types of complementarities. Explicit coordination at nodes between building blocks can be dispensed with to some extent due to their modular properties, which is why modularity is considered a necessary condition for the emergence of an ecosystem. Jacobides defines modularity as separability along a production and/or consumption chain in terms of parts that can be assembled under certain conditions. Furthermore, Kapoor and Adner describe a trade-off between performance advantage and technological challenges on the part of complementarities and focal companies. They also mention possible risks of an innovation ecosystem, which can lead to modifications of performance expectations.
Peltoniemi & Vuori 2004; Peltoniemi 2004; Peltoniemi 2006Interactions within the ecosystemPeltoniemi et al. propose four main characteristics – self-organization, emerging potential for creativity, evolution, adaptation – of an ecosystem using the evolution, systems and complexity theories and add three types of interactions: 1) selection, development as well as conscious decisions, 2) limited knowledge, 3) networking and loops through feedbacks. Thus, an ecosystem consists of different organizations that are connected by cooperation and competition – which can also run in parallel.
den Hartigh & van Asseldonk 2004;
den Hartigh et al. 2006;
Anggraeni et al. 2007
Adner 2017.
Views on structural levels Activity levelsAccording to Hartigh et al. the boundaries of an ecosystem should be variably defined by the positions of the different actors. They transfer the criteria for measuring the state of an ecosystem of Iansiti et al. from the meso level to the company level. Further, they define four main perspectives: ecosystem as network, as firm, as performance and as governance. These four perspectives are developed through comparisons between social network theory, biological ecosystem theory, complex adaptive systems theory, and business ecosystem theory. At all levels of perspective, role type, corporate strategy, network structure and dynamics, and performance are considered. We complete this class with Adner (2017), who defines an ecosystem from the perspective of mutual relationship activities – “ecosystem-as-structure”. He contrasts this perspective with the one that views an ecosystem from the perspective of the actor and its connections to other actors based on their network and platform affiliation – “ecosystem-as-affiliation”. The different approach leads to the identification of other elements (activities, actor, positions, connections) as relevant for the realization of the value proposition. In the structural approach, the relationships between companies are derived from the requirements for mutual agreement that determine positions in the overall value concept.

Other authors deal with the definition of platforms that enable the organization of ecosystems or with the description of the activities of ecosystems, e.g. Tiwana in his article ‘Evolutionary Competition in Platform Ecosystems’ from 2005. Ecosystems have also been considered and defined from strategic perspectives. We agree with Rong and Shi that such descriptions do not lead to basic characteristics of ecosystems, but to their possible models.

So far, we have been able to acquire two pieces of knowledge. On the one hand, we now know the definition of a system, which gives analyses of an ecosystem a structuring framework. On the other hand, we reduced definitions of different authors to common features, so that we can now compare all explored features of an ecosystem with those of a CAS and we do not run the risk of comparing a specialized type of ecosystem with a CAS.

To enable a comparison between an ecosystem and a CAS, we next address the description of a CAS. The comparison is followed by the conclusion, which highlights the advantages of considering an ecosystem as a CAS.

What Does the Sum of All This Add Up to?

Before applying the methods of a CAS to ecosystems, we must first show that ecosystems satisfy all the conditions of a CAS. For this purpose, we compare the characteristics of an ecosystem mentioned in various scientific articles with those of a CAS.

Holland (1995, 2014) determines basic characteristics of complex systems and their adaptive elements, so we will use his description of a CAS to contrast with ecosystems.

What does “complex” mean?

A complicated system is not the same as a complex system, so that the terms complicated and complex are by no means synonyms. We say a subject is complicated when we want to express that it has a high degree of difficulty and that we have to spend a lot of time to break it down into its component parts in order to fully penetrate it. In contrast, we call a content complex when it is multi-layered, so that breaking it down into its individual parts does not lead to an understanding of the whole. It is precisely the interconnectedness of the individual sub-units that shapes the whole system, which cannot be grasped by separate individual parts. This phenomenon, that ‘the whole is more than the sum of its parts’, is also called emergence.

According to Holland, complex systems exhibit the following :       

Since ecosystems fulfil all the requirements for being considered complex systems, we will now investigate whether the properties and mechanisms of complex, adaptive systems can also be found in ecosystems.

Are Ecosystems Complex and Adaptive?

According to Holland, a CAS has seven basic capabilities – the four properties aggregation, non-linearity, flows and diversity and the three mechanisms tagging, internal models and building blocks. The following sequence emphasizes the interrelationship between the capabilities (p. 10 ff.).

What Benefit Do We Derive from This?

Reality often shows that traditional economic models do not accurately reflect the course of interactions between firms, as their individual and collective behavior in the real world is very difficult to predict. Moreover, traditional economic models are based on differential equations and partial differential equations, which make it difficult to analyze very heterogeneous populations. In contrast, research on complex adaptive systems provides a way for economics to model and analyze both heterogeneity and possible irrational behavior. Using models that describe domains of complex adaptive systems, economists can simulate ecosystems that evolve over time and in which individuals may change their preferences based on experience and interactions with others.

Figure 2 Business Ecosystems Described from the CAS Perspective.

The change from the perspective of level 1 – individual organizations without interaction in the form of business ecosystems – to level 2 – organizations with interaction in the form of business ecosystems – takes place according to the theory of CAS through emergence

Basis for Framework Conditions, Change of Perspective and Interdisciplinary Work

The greatest benefit of describing an ecosystem as a CAS lies in the fact that relevant objects are identified, classified, and related to each other for research questions, and that various perspectives such as system hierarchy, system interior and exterior are determined. This creates basic frameworks for their exploration and enables a transition from qualitative to quantitative measurements. If internal models of a firm are interpreted as business models, the theory of complex adaptive systems could serve us as a guide on how to adapt business models depending on the stage of development of the ecosystem and how to prioritize certain actions – we would have a kind of behavioral guide for firms in an ecosystem. Furthermore, by considering an ecosystem as a CAS, we obtain a uniform vocabulary instead of the diversity of definitions of the respective authors; this also makes it possible to compare CAS from other scientific disciplines with ecosystems. Thus, models from biology, thermodynamics, information theory (cf. Stoica-Klüver et al. 2009), and many other fields can be applied to an ecosystem, providing the CC Ecosystem with additional artifacts in the DSR domain. For investigative purposes, it is important to remember that the sensitivity of a system to its initial conditions makes predictions difficult because a slight change in an organization’s past behavior or decisions can lead to completely different conditions in the future.

Simulation of Agent- and Network-Based Models

All complex adaptive systems function within the framework of certain rules and constraints. Both individual firms and multiple firms collectively can be described as complex adaptive systems with their own rules and constraints. Sugarscape is an agent-based simulation by Epstein and Axtell that models individuals, their environment, and their rules for interacting with each other and their environment. Models like Sugarscape show that changes in rules and constraints change the nature of emergence, or collective behavior. While the way individual agents respond to rules and constraints is difficult to predict, it is possible to change their behavior in such a way that the emergent behavior of the collective is directed toward achieving the goals of individuals as well as multiple entities.

By giving agents the ability to trade, for example using credit-assignment, networks and network flows emerge both between agents and between agents and their environment. Changing the absolute and relative distribution of credits along such networks changes the distribution of agents and can also lead to the occurrence of migration behavior. Using (multi-)agent-based and complex network models, it is possible to find correlations – though not causalities – between rule and constraint changes and observable events. The resulting ecosystem of rules, constraints, and action evolution interacts at different levels, resulting in new behaviors that either improve or degrade the firm’s performance against its goals. Regular assessments provide the organization with the data it needs to identify and understand its level of performance against its goals. This information can be used to update rules and constraints to steer the resulting behaviors in a more positive direction.

Cluster Concentration of Firms as Geographical Distribution

The CAS perspective dictates that companies be studied in their context. A promising context is provided by the formation of clusters. In this sense, clusters describe the concentration of firms in a certain area. With the help of cluster models in particular, it is possible to trace the outflow and inflow or migration behavior, the closure of companies and the formation of new markets within the ecosystem.

Our goal was to define universal boundaries within and around an ecosystem so that we can identify, model, and ultimately quantify processes and components. The alignment between CAS and the definitions of an ecosystem confirms that we can use CAS for boundary definitions, since ecosystems have the mechanisms of tagging, building blocks, and internal models as well as the properties of diversity, flows, aggregation, and non-linearity. As agents, companies possess adaptive and self-organizing behaviors and exhibit emergent characteristics in their interactions. The consequence is a high number of states and frequent state changes, which is called chaotic. Therefore, two main points need to be considered for investigation purposes. On the one hand, an ecosystem is sensitive to its initial conditions, so small variations can lead to large measurement differences. On the other hand, it is problematic to completely capture all states of an ecosystem, so that an information content about the imprecision of the ecosystem state should be considered for the investigation using information theory.


Bogers, Marcel and Sims, Jonathan and West, Joel. (January 15, 2019). What Is an Ecosystem? Incorporating 25 Years of Ecosystem Research. Academy of Management Proceedings, Vol. 2019, No. 1, Available at SSRN: or

Adner, R. (2017). Ecosystem as Structure: An Actionable Construct for Strategy. Journal of Management, 43(1), 39–58.

Aslaksen, E. (Hrsg.) (2008). Designing complex systems: foundations of design in the functional domain. 6000 Broken Sound Parkway NW.

Frey, T., Bossert, M. (Hrsg.) (2008). Signal- und Systemtheorie. 2., korrigierte Auflage. Wiesbaden.

Imboden, D., Koch, S., (Hrsg.) (2003). Systemanalyse Einführung in die mathematische Modellierung natürlicher Systeme. 3., korrigierter Nachdruck der 1. Auflage. Berlin, Heidelberg.

Ljung, L., (Hrsg.) (1998). Systemidentification: Theory for the user. 2. Auflage. New Jersey.

March, S., Smith, G.. (1995). Design and Natural Science Research on Information Technology. Decision Support Systems. 15. 251-266. 10.1016/0167-9236(94)00041-2.

Jacobides, MG, Cennamo, C, Gawer, A. (2018). Towards a theory of ecosystems. Strat Mgmt J. 2018; 39: 2255– 2276.

Von Bertalanffy, L., (1968). General System Theory: Foundations, Development, Applications. 2. Auflage. New York.

Rong, K., Shi, Y., (2015). Business Ecosystems Constructs, Configurations, and the Nurturing Process. New York.

Anggraeni, Elisa & Hartigh, Erik & Zegveld, Marc. (2007). Business ecosystem as a perspective for studying the relations between firms and their business networks.

Hartigh, Erik & Visscher, W. & Tol, M. & Salas, A.J.. (2013). Measuring the health of a business ecosystem. Software Ecosystems: Analyzing and Managing Business Networks in the Software Industry. 221-246. 10.4337/9781781955628.00020.

Peltoniemi, M., & Vuori, E. (2008). Business Ecosystem as the New Approach to Complex Adaptive Business Environments.

Peltoniemi, Mirva. (2005). Business ecosystem: a conceptual model of an organisation population from the perspectives of complexity and evolution.

Iansiti, Marco & Levien, Roy. (2002). The New Operational Dynamics of Business Ecosystems: Implications for Policy, Operations and Technology Strategy.

Moore, James. (1999). Predators and Prey: A New Ecology of Competition. Harvard business review. 71. 75-86.

Adner, R., & Kapoor, R. (2010). Value Creation in Innovation Ecosystems: How the Structure of Technological Interdependence Affects Firm Performance in New Technology Generations. Strategic Management Journal, 31 (3), 306-333.

Tiwana, Amrit. (2015). Evolutionary Competition in Platform Ecosystems. Information Systems Research. 26. 150424094525000. 10.1287/isre.2015.0573.

Holland, H. J., (Hrsg.) (2014). Complexity A Very Short Introduction. Oxford.

Holland, H. J., (Hrsg.) (1995). Hidden order: how adaption builds complexity. Oxford.

Nordhaus, W., D., (2011). The Economics of Tail Events with an Application to Climate Change. Review of Environmental Economics and Policy 2011 5:2, 240-257

Stoica-Klüver, C., Külver, J., Schmidt, J., (Hrsg.) (2009). Modellierung komplexer Prozesse durch naturanaloge Verfahren. Wiesbaden.

Epstein, Joshua M.; Axtell, Robert (Hrsg.) (1996). Growing artificial societies: social science from the bottom up. Washington, D.C.

Starfield, A.M., Bleloch A.L.,(Hrsg.) (1986). Building Models for Conservation and Wildlife Management. Macmillan Publishing Company, New York, and Collier Macmillan Publishers, London.

Dennis Vetterling
Ivana Vukadinovic
Latest posts by Ivana Vukadinovic (see all)

    Was sind Deine Erfahrungen mit dem Thema? (Kommentieren geht auch ohne Anmeldung oder Einloggen; einfach kommentieren, auf Freigabe warten und fertig!)