Complexity Economics

Author: Joeri Schasfoort | 18th September 2017
Scientific review will follow soon

1. Core elements

Complexity economics is the study of economic systems as complex systems. Complex systems are systems which consist of interacting individuals that change their actions and strategies in response to the outcome they mutually create (Arthur 2013). In contrast to the classical study of economic equilibria, complexity economists study the emergence of structures and the unfolding of patterns in the economy (Arthur 1999). Especially since the 2008 financial crisis, there has been increasing interest in using ideas from complexity theory to make sense of economic and financial markets (Battiston et al 2016). This increased interest was sparked because mainstream equilibrium models failed to predict it.

2. Terms, analysis, conception of economy

Complexity economists view the economy as a large complex system which consists of, belongs to, and overlap with interrelated complex systems.

Aggregate economic patterns such as economic growth and inflation are classified as emergent phenomena as they emerge from interactions of heterogeneous agents with heterogeneous expectations (Kirman 2016). These agents are only bounded rational (Simon 1972). This means that agent rationality is limited by the tractability of the decision problem, the cognitive limitations of their minds, and the time available to make the decision (Simon 1991).

When studying emergent phenomena, complexity economists such as Arthur (2013) argue that non-equilibrium is the natural state of the economy. The economy is always in a state of flux, constantly evolving and changing. There are two main reasons for this. One is fundamental uncertainty, the other is technological innovation.

The concept of (fundamental) uncertainty was introduced in economics by Knight (1921) and Keynes (1921, 1936 and 1937). They felt a distinction should be made between risk and uncertainty. In the case of risk, all possible future events or consequences of an action or decision are known. However, the probability that this event will actually materialize is unknown. Still, there are many events people simply do not know about in advance. In these situations, probability calculus has no sound foundation.

To deal with uncertainty, economic agents try to make sense of problems by surmising, making guesses, using past knowledge and experience (Arthur 2013). As a consequence, agents continually update their internal decision-making model, which means they constantly adapt or discard and replace the actions or strategies based on their experience as they explore. As a consequence, the economy is permanently in disruptive motion as agents explore, learn, and adapt.

Technological innovation is the other important contributor to the economic system’s state of permanent flux. The nature of innovation is such that technological development enables further technological development (Arthur 2013). It follows that a novel technology is not just a one-time disruption to equilibrium. Instead, it is a permanent ongoing generator and demander of further technologies that themselves generate and demand still further technologies (Arthur, 2009). Thus, technological innovation also contributes to the state of flux, be it somewhat slower than uncertainty.

There are situations where equilibrium assumptions provide useful predictions and there are situations where they can never provide useful predictions (Farmer & Geanakoplos 2008). Even if the economic system of interest approaches a state of equilibrium, Arthur (2006) argues that the static equilibrium models suffer from two more important problems: they cannot easily resolve among multiple steady states and cannot easily model individuals’ choices of expectations.

In the presence of positive feedback or increasing returns, there are often multiple equilibria. While equilibrium economics can identify multiple equilibria, it cannot tell us how one of these equilibria comes to be chosen. The equilibrium we end up with depends on the path towards that equilibrium. In other words, it is path-dependent. If a system is chaotic, tiny changes in initial conditions might even cause the system to end up in a radically different steady state (Li and Yorke 1975). What is more, once a system ends up in a steady state, it might not be straightforward to move to another steady state. It might be so resilient to changes that it takes considerable shocks to move to another regime. This is what we call a ‘lock-in.’ On the other hand, if a system’s resilience is decreasing, it might reach a tipping point and suddenly change behaviour or move to another regime (Battiston et al. 2016). Financial markets and economies have historically exhibited sudden and largely unforeseen collapses, at a systemic scale. Such “phase transitions” may in some cases have been triggered by unpredictable stochastic events. More often, however, there have been endogenous underlying processes at work (Battiston et al. 2016).

Also, self-referential expectations (economic outcomes depend on the expectations of agents today) contribute to the need for out-of-equilibrium analysis. Largely, because they cause nonlinear system dynamics. Arthur (1994) shows this in the famous El-Farol bar model. In this model, agents determine whether or not to go to a bar. Their decision to go depends on their expectations about how crowded the bar is. If they expect it to be crowded they will stay home and vice versa. Agents learn about the actual crowdedness of the bar the day after –even if they stayed home. As you might expect no equilibrium will emerge as bar attendance will fluctuate because of the negative relationship between expectations and attendance. Not only assuming equilibrium does not hold in this case, it would fail to predict the fluctuating bar attendance.

3. Ontology

As argued before, complexity economics is a reaction to mainstream economics in which equilibrium is the natural state of the economy. Therefore, continuous change is the central economic problem of complexity economics. Uncertainty and technological change are the crucial variables which explain the economies’ permanent deviation from equilibrium.

The smallest parts of complex systems are agents. Economic agents can be human beings or institutions such as firms, banks or governments. Agent decisions are driven by humans or groups of humans.

Agents generally do not optimize (e.g. utility) in the standard sense (Arthur 2010). Rather, people engage in cognitive processes such as social comparison, imitation and repetitive behaviour (habits) so as to efficiently use their limited cognitive resources (Jager et al 2000). Still, complexity economists generally recognize that even their more complicated cognitive models are a brutal simplification of reality. To make sure the simplification is appropriate for the problem at hand, cognitive models should be validated against empirical data (Jager 2000).

That being said, agent decision-making depends on their surroundings. Their interactions are shaped by the decisions of other agents as well as institutions such as the rule of law, culture, and markets. They are also bounded by environmental constraints. Together, institutions and individuals form a complex system. At the same time, the system shapes the institutional structure and human decisions. These elements on their own cannot explain economic phenomena. Economics is more than the sum of its parts.

Furthermore, complexity economists often explicitly model time. They use discrete time –often the case in computational models– or continuous time –often the case in analytical models. Time plays an important role in complexity economics, due to path-dependence. The present and past states of economic systems rely on its past. Time can, therefore, be said to be interpreted as static (Čapek, M. 1976). Viewed this way, the economy becomes a system that evolves procedurally in a series of events; it becomes algorithmic (Arthur 2013).

Complex economic systems are always overlapping with other complex systems but not always in a hierarchical way. Rather, these overlapping systems can be described as a Panarchy (Holling 2001): “the structure in which systems are interlinked in continual adaptive cycles of growth, accumulation, restructuring, and renewal.” Complex systems have complex systems above and below them. At the same time, they are part of multiple overlapping complex systems.

For example, the stock market is a complex system which consists of investors who trade stocks using the stock exchange platform. These investors are often not individuals but institutions. These institutions, in turn, are a complex system which is formed by different stakeholders such as employees, management, and shareholders. On the other side, the stock market is part of the larger economic ecosystem. Finally, the stock market overlaps with several other systems. The financial press, for example, is simultaneously an important determinant of the stock market while being part of the media ecosystem.

4. Epistemology

Complexity economics is quite general, it applies a certain mode of thought –viewing the world as a complex system- and applies this to all sorts of economic problems. In practice, this world can only partly be observed by humans. In a complex adaptive system, the observer, in this case the economist, cannot be fully separated from the system.

Trying to make sense of the world, scientists observe the natural world then encode it into another system that is of their making or choosing, which we can call a model. The model can then be manipulated with the objective of trying to replicate the causal relationships in the natural world. Epstein’s (2006) calls this a generative social sciences vision. In this vision, science is first and foremost explanatory and the role of the complexity economist is to generate processes of interest computationally. Finally, the model must then be decoded –back to the natural world- in order to check its success or failure in representing the causal event (Mikulecky 2001).

However, according to Mikulecky (2001), complexity scientists must go beyond this method and recognize that true complexity as the property of a real-world system means that no model is able to adequately capture all its properties.

5. Methodology

Recognizing that their models never capture all real-world properties, economic enquiries start with some observed phenomena and an observer’s preconceptions of the causal mechanisms that lie behind these observations. To understand why the phenomenon occurs, complexity economists develop theories which they formalize using mathematical models. These models yield several hypotheses which can be tested. Hypotheses should then be taken to empirical data. Using the data, scientists will try to reject the theory. If a theory has been rigorously tested and not rejected it becomes a plausible representation of reality until it is rejected.

Preferably, complexity economists use mathematical models to formalize theories. When done right, mathematical models are unambiguous and they expose internal inconsistencies and implied predictions. So far, this methodology is similar to the way many non-complexity economists operate. However, in contrast to mainstream economics, complexity economists use either non-linear dynamics or agent-based models because both techniques can represent behaviour out of equilibrium.

For example, to explain key stylized facts observed in financial markets, such as stationary returns, excess kurtosis and volatility persistence, Franke and Westerhoff (2012) model bounded rational investors who switch between trend-following and fundamentalist strategies depending on which strategy is more profitable at that point in time. In contrast, it is well known that standard (consumption based) asset pricing models under rational expectations are at odds with these basic facts (Adam et al. 2016). According to the efficient market hypothesis, stock prices should follow a random walk (Lo 2004).

In macroeconomics, Steve Keen’s (1995) nonlinear dynamics model predicted a long period of apparent stability (the great moderation) and the following crisis (the great financial crisis of 2008) where equilibrium models predicted that macroeconomic systems tend towards equilibrium. The crucial difference is that nonlinear dynamics models (in macroeconomics also known as Stock-Flow Consistent models (Godley and Lavoie 2006) could account for the nonlinear dynamics brought on by stocks of debt, see Bezemer (2010) for an extended discussion.

Together with standard equilibrium-based models, the nonlinear dynamics models are sometimes referred to as equation based models (EBM) (Railsback 2001) because they consist of top-down equations which represent aggregate flows. Such aggregate flows are emergent properties of the system. These patterns are a result of individual interactions but they are not necessarily equal to individual behaviour (Goldstein 1999).

However, if these aggregate relationships change over time, understanding them can be especially problematic. It means that they cannot be reliably estimated from past data. This criticism was at the core of the Lucas critique (1976). Complexity economists would argue that it applies to all equation-based models, including those with representative agent micro foundations (Kirman 1992). In such cases, it can be appropriate to model the individual interactions directly, enter agent-based modelling.

Agent-based models are simulated systems in which aggregate outcomes are determined by the social interactions of explicitly modelled agents with limited and local knowledge (Bowles et al. 2017). In general, these bottom-up models are bigger and more complicated than their equation-based counterparts. If the behaviour of aggregate system relationships is not fully understood, the additional effort of building an agent-based model might be necessary.

Whichever type of model a complexity economist uses, the predictions it hypotheses yield should then be validated using empirical data. Ideally, models should be able to account for multiple patterns at the same time, see Grimm et al. (2005).

6. Ideology and political goals

Complexity economists seek to better understand the behaviour of our complex and uncertain world. Often, this translates to a desire to improve the workings of our economic system. Positive economics is in principle independent of any particular ethical position or normative judgments (Friedman 1956). If a difference between positive and normative economics can be made like this, complexity economist are generally positive economists. That is, when they try to understand economic systems, they aim at being as objective as possible and try to prove or disprove relationships.

7. Current debates and analyses

As a new school of thought, prominent complexity economists are mostly preoccupied trying to convince economists to expand their analysis beyond a rational expectations, representative agent, and equilibrium focus. That being said, there are some interesting debates among complexity economists who are at the forefront of developing agent-based models.

Agent-based modellers generally balance between two rules (Sun et al. 2016): Keep It Simple Stupid (KISS, Axelrod 1997), and the Keep It Descriptive Stupid (KIDS, Edmonds & Moss 2004). The first argues that a model should be kept as simple as possible. The second argues that the model should be detailed enough to model the richness of target systems. While complexity models can model highly complicated cognitive agents (Sun 2006), complexity economists generally recognize that real human beings are far more complicated than their models allow.

Another debate was inspired by the frustrations with difficulties in describing and replicating agent-based models. Confronted with the lack of a protocol for agent-based modelling descriptions, Volker Grimm and others (2010) developed one. It is called the Overview, Design and Details (ODD) protocol. While many modellers have adopted it, especially in ecology, there is still a lack of uniformity in the description of agent-based models in economics.

As a relatively new tool, there is still no consensus on how agent-based models should be validated. It is generally agreed that, agent-based models should be able to replicate some key stylized facts. Preferably, they should be able to replicate multiple of these key patterns at the same time (Grimm et al. 2005). Yet, Guerini and Moneta (2017) note that models which incorporate different causal structures may replicate the same stylized facts. They, therefore, propose a method to focus only on representing causal structures among aggregate variables of the ABM and test whether they significantly differ from the causal structures that can be found in the real world from observed aggregate variables.

8. Delineation: subschools, other disciplines, other economic theories

Viewing the economy as a complex system, complexity economics is methodologically focused. This makes it compatible with most other non-mainstream schools of economic thought, especially Austrian, behavioural, ecological, evolutionary, institutional, and post-Keynesian economics. It is not unheard of that complexity economists use cognitive rules from behavioural, neo-classical and post-Keynesian economics in an evolutionary framework. As complexity economics was inspired by the broader complexity science movement, it also draws inspiration from other disciplines such as biology, ecology, physics, and mathematics.

From the perspective of complexity economics, behavioural, ecological, evolutionary, and institutional economics focus on specific aspects of complex adaptive systems. Behavioral economists focus on the agents’ decision-making process. Institutional economists focus on the institutions that facilitate and shape their decisions. Evolutionary economists study the selection mechanisms that give rise to both behaviour and institutions. Finally, ecological economists study the sustainability of the complex system and its relationship with other non-economic complex systems. Austrian and post-Keynesian economics differ from these in their broader assumptions about the economic system.

Still, these are often compatible with complexity economics. Early on, Austrian economists viewed the economy as a complex adaptive system. According to Veetil and White (2017): “Austrian macroeconomists of the interwar period saw the economy as a complex adaptive system, in which macroeconomic variables emerge from the interaction between millions of purposefully acting agents.” Furthermore as Bowles, Kirman and Sethi (2017) explain: “Friedrich Hayek is known for his vision of the market economy as an information processing system characterized by spontaneous order: the emergence of coherence through the independent actions of large numbers of individuals, each with limited and local knowledge, coordinated by prices that arise from decentralized processes of competition.” Finally, the New Austrian (also called Neo-Mengerian) paradigm emphasizes the importance of non-equilibrium and emergent processes in explaining the social world (Salter 2017).

Likewise, post-Keynesians emphasize fundamental uncertainty, the importance of institutions, decision heuristics and deviations from equilibrium in the form of instability (Aboobaker, Köhler, Prante and Tarne 2016). The staple post-Keynesian modelling technique of stock-flow consistent modelling (Godley and Lavoie 2006) can be seen as dynamical systems mathematics applied to monetary macroeconomics. Increasingly this technique is being combined with agent-based modelling, see for example Caiani et al. (2016).

9. Delineation from the mainstream

Complexity economics was partially developed in contrast to the prevailing neo-classical economics paradigm and especially the mainstream reliance on the concept of equilibrium. The main difference between complexity economics and the mainstream is, therefore, the focus on equilibria–static patterns that call for no further behavioural adjustments. Complexity economics portrays the economy not as deterministic, predictable, and mechanistic, but as process dependent, organic, and always evolving (Arthur 1999). Still, equilibrium economics is not discarded in its entirety (Farmer & Geanakoplos, 2009). From the perspective of complexity economics, equilibrium economics is a special case of non-equilibrium and hence complexity economics.

Complexity economists have especially been vocal critics of the neo-classical / new-Keynesian rational expectations hypothesis, stating that it is unrealistic and potentially invalidates model outcomes. On the other hand, complexity economists try to explain observed phenomena by showing how a population of cognitively plausible agents, interacting under plausible rules, could actually arrive at these phenomena on time scales of interest (Epstein 2006). Or as Epstein (1999) puts it: “if you didn’t grow it, you didn’t show it.” This is very different to the mainstream economics equilibrium models which are the dominant way to explain economic phenomena. Using these equilibrium (?) models, it is sufficient to demonstrate that the observed phenomena can be the Nash-equilibrium of some game (Arthur 1999).

10. Institutions


Complexity economists publish in both ‘mainstream’ and ‘heterodox’ journals (Heise 2016), including highly ranked journals such as Nature, Science, The American Economic Review, Econometrica, and the Journal of Economic Perspectives. Furthermore, there are a few journals which either focus on complexity and agent-based modelling or regularly publish such studies. These include:

  1. Advances in complex systems,
  2. Complexity,
  3. Computational economics,
  4. Journal of economic behavior and organization,
  5. Journal for artificial societies and social simulation,
  6. Journal of evolutionary economics,
  7. Quantitative finance.

Think tanks

There are several think tanks and university departments which are specialized in complexity economics. The most famous of these is the Santa Fé Institute. University departments which are actively involved in economic complexity include:

  1. Institute for New Economic Thinking (INET) at the Oxford Martin School
  2. Institute of Complex Systems (ISC), Rome
  3. Center for the Study of Complex Systems at the University of Michigan
  4. University of Amsterdam Center for Non-Linear Dynamics in Economics and Finance (CENDEF)
  5. Groningen center for social complexity studies (GCSCS)
  6. London School of Economics (LSE) Complexity Group.


Adam, K., Marcet, A. and Nicolini, J.P., 2016. Stock market volatility and learning. The Journal of Finance, 71(1), pp.33-82.

Arthur, W. B. (1994). Inductive reasoning and bounded rationality. The American economic review, 84(2), 406-411.

Arthur, W. B. (1999). Complexity and the economy. science, 284(5411), 107-109.

Arthur, W. B. (2006). Out-of-equilibrium economics and agent-based modeling. Handbook of computational economics, 2, 1551-1564.

Arthur, W. (2010). COMPLEXITY, THE SANTA FE APPROACH, AND NON-EQUILIBRIUM ECONOMICS. History of Economic Ideas, 18(2), 149-166. Retrieved from

Arthur, W. B. (2013) Complexity Economics: A Different Framework for Economic Thought.

Axelrod, R. M. (1997). The complexity of cooperation: Agent-based models of competition and collaboration. Princeton University Press.

Battiston, S., Farmer, J. D., Flache, A., Garlaschelli, D., Haldane, A. G., Heesterbeek, H., ... & Scheffer, M. (2016). Complexity theory and financial regulation. Science, 351(6275), 818-819.

Bowles, Samuel, Alan Kirman, and Rajiv Sethi. 2017. "Retrospectives: Friedrich Hayek and the Market Algorithm." Journal of Economic Perspectives, 31(3): 215-30.

Čapek, M. (1976). The concepts of space and time.

Caiani, A., Godin, A., Caverzasi, E., Gallegati, M., Kinsella, S. and Stiglitz, J.E., 2016. Agent based-stock flow consistent macroeconomics: Towards a benchmark model. Journal of Economic Dynamics and Control, 69, pp.375-408.

Edmonds, B., & Moss, S. (2004, July). From KISS to KIDS–an ‘anti-simplistic’modelling approach. In International Workshop on Multi-Agent Systems and Agent-Based Simulation (pp. 130-144). Springer Berlin Heidelberg.

Epstein, J. M. (2006). Generative social science: Studies in agent-based computational modeling. Princeton University Press.

Epstein, J. M. (2006). Remarks on the foundations of agent-based generative social science. Handbook of computational economics, 2, 1585-1604.

Farmer, J. D., & Geanakoplos, J. (2009). The virtues and vices of equilibrium and the future of financial economics. Complexity, 14(3), 11-38.

Reiner Franke and Frank Westerhoff. Structural stochastic volatility in asset pricing dynamics: Estimation and model contest. Journal of Economic Dynamcis and Control, 36:1193–1211, 2012.

Friedman, M. (1953). The methodology of positive economics.

Godley, Wynne, and Marc Lavoie. Monetary economics: an integrated approach to credit, money, income, production and wealth. Springer, 2006.

Goldstein, J., 1999. Emergence as a construct: History and issues. Emergence, 1(1), pp.49-72.

Grimm, V., Revilla, E., Berger, U., Jeltsch, F., Mooij, W. M., Railsback, S. F., ... & DeAngelis, D. L. (2005). Pattern-oriented modeling of agent-based complex systems: lessons from ecology. science, 310(5750), 987-991.

Grimm, V., Berger, U., DeAngelis, D. L., Polhill, J. G., Giske, J., & Railsback, S. F. (2010). The ODD protocol: a review and first update. Ecological modelling, 221(23), 2760-2768.

Mattia Guerini and Alessio Moneta. A method for agent-based models validation". In: Journal of Economic Dynamics and Control (2017).

Heise, A. (2016). Whither economic complexity?.

Holling, C. S. (2001). Understanding the complexity of economic, ecological, and social systems. Ecosystems, 4(5), 390-405.

Jager, W., Janssen, M.A., De Vries, H.J.M., De Greef, J. and Vlek, C.A.J., 2000. Behaviour in commons dilemmas: Homo economicus and Homo psychologicus in an ecological-economic model. Ecological economics, 35(3), pp.357-379.

Keen, S., 1995. Finance and economic breakdown: modeling Minsky’s “financial instability hypothesis”. Journal of Post Keynesian Economics, 17(4), pp.607-635.

Keynes, J.M. (1921), A Treatise on Probability, The Collected Writings of John Maynard Keynes, Vol. VIII, London

Keynes, J.M. (1936), The General Theory of Employment, Interest and Money, The Collected Writings of John Maynard Keynes, Vol. VII, London, Chapter 12 in particular

Keynes, J.M. (1937), “The General Theory of Employment”, reprinted in The Collected Writings of John Maynard Keynes, Vol. XIV, pp. 109-124

Kirman, A.P., 1992. Whom or what does the representative individual represent?. The Journal of Economic Perspectives, 6(2), pp.117-136.

Kirman A., 2016, Complexity and economic policy,

Kirman, Alan. "Ants and nonoptimal self-organization: Lessons for macroeconomics." Macroeconomic Dynamics 20.2 (2016): 601-621.

Knight, F. H. (1921): Risk, Uncertainty and Profit. Boston: Houghton Mifflin

Lo, A.W., 2004. The adaptive markets hypothesis. The Journal of Portfolio Management, 30(5), pp.15-29.

Li, T.Y. and Yorke, J.A., 1975. Period three implies chaos. The American Mathematical Monthly, 82(10), pp.985-992.

Mikulecky, D.C., 2001. The emergence of complexity: science coming of age or science growing old?. Computers & chemistry, 25(4), pp.341-348.

Railsback, S.F., 2001. Concepts from complex adaptive systems as a framework for individual-based modelling. Ecological modelling, 139(1), pp.47-62.Jager, W. (2000). Modelling consumer behaviour s.n.

Salter, A.W. Rev Austrian Econ (2017) 30: 39. doi:10.1007/s11138-016-0350-3

Simon, H.A., 1972. Theories of bounded rationality. Decision and organization, 1(1), pp.161-176.Schulze, J. Müllera B., Groeneveld, J., & Grimm, V. (2017), Agent-Based Modelling of Social-Ecological Systems: Achievements, Challenges, and a Way Forward, Journal of artificial societies and social simulation 20 (2).

Simon, H. A. (1991). Bounded rationality and organizational learning. Organization science, 2(1), 125-134.

Sun, R. (2006). Prolegomena to integrating cognitive modeling and social simulation. Cognition and multi-agent interaction: from cognitive modeling to social simulation, 3-26.

Sun, Z., Lorscheid, I., Millington, J. D., Lauf, S., Magliocca, N. R., Groeneveld, J., ... & Buchmann, C. M. (2016). Simple or complicated agent-based models? A complicated issue. Environmental Modelling & Software, 86, 56-67.

Veetil, V.P. & White, L.H. Rev Austrian Econ (2017) 30: 19. doi:10.1007/s11138-016-0354-z


Assigned course modules

Title Provider Start Level
Thinking Complexity Toulouse Business School self paced beginner
Complexity Economics - self paced advanced
Emergence Theory - self paced beginner
Introduction to Dynamical Systems and Chaos Santa Fe Institute currently not in session beginner
Nonlinear Dynamics: Mathematical and Computational Approaches Santa Fe Institute currently not in session advanced
Complex adaptive systems - self paced advanced
Network Theory Introduction - self paced advanced
Introduction to Complexity Santa Fe Institute always beginner

Organisations and links


This project is brought to you by the Network for Pluralist Economics (Netzwerk Plurale Ökonomik e.V.).  It is committed to diversity and independence and is dependent on donations from people like you. Regular or one-off donations would be greatly appreciated.