I have already made some comments on the philosophy of econophysics and some comments on monetary economics from an econophysics perspective. Now I want to provide some commentary from an economist on econophysics, based on the work Economia da Complexidade. To be honest, much of what is written here consists mainly of excerpts taken from the book itself, translated and reorganized in order to present what I understand to be the central idea of the entire material in a coherent way.

Introduction¶

Econophysics is a field of study aimed at relating or explaining economic phenomena with the aid of techniques from Physics. The techniques typically used for this purpose involve nonlinear dynamics, stochastic processes, and uncertainties. Within Probability Theory, a stochastic process is a family of random variables representing the evolution of a system of values over time. It is the probabilistic counterpart of a deterministic process.

Instead of a process with a single path of evolution, a stochastic process has inherent indeterminacy. With small variations, there are multiple—sometimes infinitely many—directions in which the process can evolve away from a given condition. When the natural system under study is nature itself, and encoding and decoding are performed using Mathematics, it is common to say that what is being done is Physics.

A relevant point for understanding what a physicist does is to compare them with an economist. In orthodox economics, the “rule” is atomistic reductionism, in which reality must be explained in terms of a representative rational agent. Orthodox economists assume that the aggregate behavior of a system is identical to the sum of the effects of each individual cause.

In turn, the physicist uses, as a “principle,” interactive reductionism; that is, they do not attempt to describe a system’s agent in isolation, but rather aim to describe the interactions of an agent with others. For a physicist, the interaction between agents results in a statistical description of the system’s aggregate behavior.

Economists labeled as heterodox engage in Econophysics when they incorporate the idea that the economy’s resulting performance emerges from the interactions among its diverse components: the psychologies of economic agents, institutions, adaptations, innovations, disruptions, and so on. They seek a transdisciplinary perspective.

The theoretical avant-garde of economists aims to develop a Social Physics in order to expand socioeconomic and political thinking, including not only competitive forces but also exchanges of ideas, information, and pressures for social status, in order to better explain collective human behavior. It seeks to explain how social interactions generate the dynamic self-organization of the economy as a complex system—that is, to understand how from the social interactions of system components, from individual agents to the community, the market, and the state, a networked society emerges.

Making Sense of Chaos¶

J. Doyne Farmer is an American entrepreneur, complex systems scientist, and author of the book Making Sense of Chaos: A Better Economics. Although it took him ten years to write, in it the physicist partially presents his current research in economics, including Agent-Based Modeling (ABM), financial instability, and technological progress.

The economic impact of the COVID pandemic combined three exceptional characteristics that made standard models inadequate:

1. The shocks were sector-specific, making it essential to model the economy with high resolution;

2. The shocks simultaneously affected both supply and demand;

3. The shocks were very strong and occurred very rapidly.

Conventional economic models passively wait for a return to equilibrium—that is, for supply and demand to become equal. It was therefore necessary to quickly build a heterodox model capable of tracking the dynamics of the shocks without assuming the existence of equilibrium. Complexity economists provided a model that achieved what is called verisimilitude: it made the most important mechanisms capable of affecting the economy realistic enough to deliver accurate forecasts. This served as a proof of concept for the usefulness of Complexity Economics—the central theme of Farmer’s book.

Complexity Economics is a transdisciplinary movement of rebellious economists and other scientists against orthodoxy. It seeks to better understand the economy by using principles entirely different from those of standard economics. The COVID model was the first to use methods of complex economics to make an accurate prediction of a major economic event, outperforming traditional models.

The core assumptions of mainstream economics do not correspond to reality, and methods based on them do not scale well from small to large problems. Moreover, they fail to take full advantage of huge advances in data and technology. The incorrect answers of standard economic theory end up justifying inaction—namely, the outdated laissez-faire approach of the free market.

Philosophical Difference¶

The diagnosis made by the physicist turned quasi-economist, J. Doyne Farmer, is that the problems faced by orthodox economics are inherent to the conceptual structure dominant within the profession. This conceptual framework, he argues, “is not suitable for dealing with large, messy, complicated, real-world problems.”

Idealist Philosophy and Materialist Philosophy are two major philosophical traditions in the Western world. They differ radically in their approaches to the nature of reality, the origin of knowledge, and the relationship between mind and matter.

Idealist Philosophy emphasizes the primacy of ideas, mind, or consciousness as the foundation of reality. For idealists, reality is ultimately mental or spiritual—not physical. According to this view, material existence depends on mental perception: objects exist only when they are perceived by a mind. Material entities are manifestations or representations of an underlying spiritual reality. Knowledge, in turn, is obtained independently of sensory experience, through reason or introspection. Certain categories of understanding are considered innate and serve to structure our experience of the world.

In contrast, Materialist Philosophy emphasizes the primacy of matter as the foundation of all reality. For materialists, consciousness and ideas are products of material processes. Matter is the fundamental substance of the universe, and all phenomena—including consciousness—arise from material interactions. Thought and consciousness are the results of the brain’s material activity. Ultimate reality is physical and can be understood through the study of material properties and interactions.

For example, Karl Marx’s view was that the economic infrastructure (relations of production and productive forces) determines the social superstructure (political, cultural, and ideological institutions). In this line of thought, materialist scientists, such Doyne Farmer, base their understanding of the universe on experimentation and empirical evidence. Reality exists independently of human perception.

Because of this epistemological divide, idealist economists place emphasis on ideas, moral and spiritual values, introspection—and on neoliberal ideology itself! In contrast, materialists emphasize economic analysis, empirical science, and the material transformation of social conditions.

This opposition is reflected in philosophical, scientific, and social thought, shaping debates about the nature of being, knowledge, and society. In the case of J. Doyne Farmer, by “theory,” he means building models of the world based on assumptions about how it actually works.

Agent-Based Modeling¶

There is a standard model for constructing economic theories that incorporates the agents’ reasoning capacity. This model, linked to standard economic theory, assigns each agent a utility function describing their preferences—and how to maximize their own utility or happiness. Equations express this mathematically.

Complexity Economics offers a completely different alternative by using ideas and methods from the science of complex systems, a transdisciplinary movement that studies emergent phenomena in complex systems. Emergent phenomena occur when the behavior of a system as a whole is qualitatively different from that of its individual or interactive components.

Complexity Economics assumes that agents are boundedly rational—that is, they make imperfect decisions and have limited reasoning capacity. Agents can learn to pursue goals, but they usually achieve them only partially. Their decisions are decentralized and uncoordinated.

According to J. Doyne Farmer, although it also uses equations, computational simulation is the flagship of Complexity Economics. Digital technology allows it to create real-world analogs with computer support. Scientists are using computer simulations to study almost everything that can be interpreted as a complex system. Traditional economists use computers to solve equations—but that is not simulation in the modern sense.

The simulations used in Complexity Economics are called Agent-Based Models (ABMs). An Agent-Based Model is a computational simulation approach used to study complex systems composed of multiple interacting agents. This model is useful across various disciplines—including Economics, Sociology, Ecology, and Computer Science—to understand how local interactions among individual agents can lead to emergent patterns at the macro level.

An ABM is a class of computational models in which the behavior of a system is simulated through the interactions of autonomous and individual agents. These agents represent people, firms, or any entity capable of making decisions based on simple rules.

• Agents: These are individual entities with specific behaviors and decision-making capabilities. Each agent operates according to a set of rules or algorithms that define its actions. They have attributes such as memory, learning ability, goals, and decision-making strategies.

• Environment: This is the space where agents interact. It can be physical, such as a national territory, or abstract, such as a network of social interactions. It has its own rules and dynamics—such as spatial constraints or available resources—that affect the agents.

• Interactions: These occur among agents and between agents and the environment, and are fundamental to ABMs. Economic exchanges, cooperation, competition, communication, and migration are examples of possible interactions.

• Behavioral rules: These determine how agents make decisions based on their internal state and the state of the environment, thereby shaping how interactions occur. They may include optimization algorithms, simple heuristics, machine learning, or logic-based models.

• Computational simulation: This involves running the model over time, allowing agents’ interactions and behaviors to evolve. Time can be represented in a discrete or continuous manner. The evolution of the system is observed through changes in the state of the agents and the environment.

In the context of ABMs, agency refers to the ability of agents to make autonomous decisions and act based on those decisions. Agency is crucial because it allows each agent to contribute to the system’s overall emergent behavior. Agents can act according to their own rules and objectives without the need for centralized coordination. They usually differ from one another in terms of attributes, behaviors, and goals, reflecting the diversity found in real systems, where heterogeneity predominates.

Although decisions are made locally by each agent, their collective interactions lead to global emergent patterns that are not easily predictable from individual behavior. Examples of ABM applications are multiplying across disciplines — in economics, sociology (e.g., information diffusion), ecology (e.g., population dynamics), and more. Therefore, ABMs are valuable tools for understanding complex systems, where emergent behavior results from the interactions of many autonomous agents.

There are many advantages to using simulations instead of solving mathematical equations. They make it possible to model the diversity of real-world participants — what these economists call heterogeneity. Standard economic theory assumes that transactions occur only when supply equals demand, a state known as equilibrium. In contrast, complex economy models consider equilibrium, when it does occur, as an emergent property.

Taken together, all these elements lead to a very different way of dealing with change. Change arises not only “from the outside” but also “from within,” when market forces act upon themselves. Standard economics has difficulty dealing with transformations that originate within the free market. Conventional financial crisis models tended to assume only exogenous shocks as causes.

Standard economic theory produces tractable models in simple environments, but this approach fails when things get complicated — the equations become too difficult to solve, and adding new features to an existing model becomes increasingly hard. As a result, when a problem grows complex, traditional economists are forced to oversimplify, leaving the “spurious” — the messy, realistic part — out. Building realistic models requires good data. ABMs are naturally well-suited to exploit the vast amounts of data now available on the trillions of transactions recorded annually.

This modeling approach studies the economy from the bottom up: macroeconomics emerges from microeconomics. Milton Friedman once argued that models should be judged solely by their predictive power, rather than by the plausibility of their assumptions. Even if those assumptions were unrealistic, economists should use them “as if” they were true.

Farmer warns that we should be deeply suspicious of models that rely on implausible “as if” assumptions. Instead, we should follow what he calls the principle of verisimilitude: models should fit the facts, and their assumptions should be plausible. Unrealistic assumptions are more likely to lead to false conclusions than plausible ones.

The principle of verisimilitude recognizes that models must include the key characteristics of the phenomena they aim to explain, but they don’t need to be literal representations of the world. By definition, models are abstractions; they don’t have to capture every detail. Models operate within a specific environment and are not applicable in another due to contextual differences. They are not a general theory. Verisimilitude simply means capturing the essential components as realistically as possible. Good models should be as simple as possible — but not too simple.

Agent-Based Models (ABMs) can be simple or complex, depending on necessity. After all, computers can easily handle details that are mathematically difficult to formulate, allowing us to include as many features as we need. Moreover, it is possible to add new features without altering existing ones, gradually increasing the realism of the world model. The transdisciplinary approach, whose ideas have significantly influenced this field, has shown that the use of practical decision-making rules is often more effective, in many contexts, than traditional optimization models.

While complexity economists remain at the margins of academic economics, their models are beginning to gain support from institutions dissatisfied with conventional tools. This includes some avant-garde Central Banks that use such models to monitor financial markets and prevent new collapses. The new science of Complexity Economics represents a radical shift in the way economics is taught. The fact that its intellectual structure is so different implies a complete reformulation of its tools, skill base, and body of knowledge — that is, a profound cultural transformation. Such a significant change is strongly resisted by the mainstream, as revolutions always are.

Predictive Capacity¶

Orthodox economists generally assume that prices are in equilibrium, defined as the point where supply equals demand. However, in reality, in most markets, transactions occur at prices where supply does not equal demand. One of the contributions of Complexity Economics is understanding how prices move when markets operate out of equilibrium. It views the economy as an evolving complex system.

Farmer sought to move away from the reductionist approach. During graduate school, he discovered Cybernetics, the precursor to complex systems theory. Although most economists share some common ground, they adopt a variety of different models and make divergent predictions. Their readers are free to choose whichever model benefits them most, based on ideology, politics, or self-interest.

The goal in the aforementioned book is to present a view of how we can build models capable of producing the best economic predictions. Surprisingly, many economists do not agree on the importance of this goal, as predictive accuracy is rarely emphasized in economic research. In this sense, the field differs greatly from research in Physics.

Farmer argues that models with weak predictive power are limited in their ability to improve our economic understanding of the world. He asks: if we cannot make reliable predictions, then how do we know whether the conceptual framework we use to think about the world and evaluate policy choices is correct?

We will never be able to predict the economy perfectly — far from it — but we can do much better than what has been done so far. Yes, prediction is difficult, but there are many examples of good forecasts. Changes in the economy fall into two basic categories: those coming from outside the economic system and those generated internally.

• Standard explanation: The explanation offered by standard laissez-faire economic models is that changes in the economy are always initiated by external influences, called exogenous shocks: random events originating outside the economy and, by definition, unpredictable.

• Alternative explanation: The alternative or heterodox explanation is that change comes from within — in technical jargon, that movement is endogenous: the economy changes on its own due to internal factors.

In reality, it is a mixture of both: sometimes the economy responds to external shocks, sometimes changes arise internally, and sometimes both types of causes act simultaneously.

Models capable of describing how things change over time are known as dynamical systems. Most dynamical systems have what is called an attractor. This attractor determines their long-term behavior, that is, the state toward which the system tends to evolve and settle into equilibrium. However, there is an attractor with a very different nature from what physicists were used to, and it was appropriately called a chaotic attractor. The motion it produces came to be known as chaos.

Chaos is characterized by two essential properties: first, sensitive dependence on initial conditions; second, endogenous motion, meaning that even if there are no external shocks, for practically any initial condition that is not an unstable equilibrium point, the system will display continuous and aperiodic movement.

From this, Farmer raises the key question: is the economy chaotic? Traditional Economic Science does not confront chaos. A problem with prevailing macroeconomic models is that they have only fixed-point attractors. Thus, if the market is left alone, it will enter a resting state and stay there forever. This implies, for orthodox economists, that all crises or changes in the economy are driven only by external events, never by the free market itself.

To explain why a market economy changes, economists have to postulate the existence of shocks capable of pushing the model away from equilibrium. They do not predict the shocks, only how the economy will relax back to equilibrium. For complexity economics, if the economy produces endogenous motion that is not a simple limit cycle or a combination of limit cycles, the hypothesis is that it is due to chaos. In models of complex economics, chaos is common. It explains how randomness emerges from order, but it is only a small subfield within complex systems.

The Economy as a Complex System¶

Because the economy is a complex adaptive system, biological concepts such as metabolism, ecology, and evolution are very useful for thinking about it. They go beyond mere metaphors, because they contain general principles that help us understand how the economy is organized and how it functions. Adam Smith, in his book The Wealth of Nations (1776), had already described the economy as a complex system. In modern terms, he saw the economy as an emergent phenomenon.

Understanding this system means thinking of the economy in terms of networks: they provide a universal language capable of describing and accounting for the operations of complex systems. Networks are one of the central ideas of Complexity Economics because they transmit behavior and accounting.

Networks identify the essential building blocks of a complex system and provide a schematic view of their interactions. The skeleton of the modern economy is a vast network of balance sheets. Each balance sheet is a list of assets and liabilities, including both physical goods and services and contracts such as money. The nodes of the network correspond to any organization with a balance sheet, explicitly or implicitly, of accounting records.

Complexity economists think of the economy schematically as an interaction between accounting and human decision-making. Accounting is represented by the network of balance sheets, and people make economic decisions. All these decisions constitute economic activity. To understand it, we need to study how human behavior interacts with the underlying network of balance sheets.

Tracing and understanding the vast and complicated interconnected balance sheets of the globalized economy is a challenge. Modeling human decision-making, both as individuals and in groups, is an even greater challenge. Hence the need to simplify the problem. The balance-sheet network is divided into parts, and each part is aggregated.

We take averages of economic activity measures for countries or regions and study their interactions. Macroeconomics studies the interactions of aggregated quantities such as GDP, unemployment, inflation and interest rates, exchange flows, within and between nations. Microeconomics studies the interactions of balance sheets on a more detailed scale, but without attempting to see the entire economy.

Conventional economics understands the balance-sheet network from the top down through the assumption that aggregates represent rational and uniform behavior of economic agents. Complexity Economics views it as woven from the bottom up because it observes heterogeneous behaviors at least at the level of the so-called institutional sectors.

Financial System¶

Farmer states that the essential function of the economy is to structure and coordinate our work and allocate resources, among which is human capital. In a prosperous economy, individuals are collectively far more effective compared to a situation in which each must survive alone. The financial system is supposed to play an essential role in guiding this collective coordination. The fundamental function of the financial system is to guide and control the economy, determining what is or is not necessary to produce. Thus, it plays a key role in processing information and providing indicators for capital-allocation decisions.

However, Farmer questions whether the main components of the financial system do anything more than act as a global casino for the wealthy themselves, while destabilizing the economy with few positive benefits for the economy as a whole.

Eugene Fama’s Efficient Market Hypothesis argues that the financial system functions perfectly, whereas J. Doyne Farmer’s Market Ecology Theory suggests that inefficiencies are inherent to the system and essential for understanding its failures. It questions whether the system’s growth reflects a real need for information processing or represents a dysfunctional deviation.

The debate in the 1990s was dominated by Eugene Fama, a professor at the University of Chicago Booth School of Business and winner of the 2013 Nobel Prize in Economics, regarded by orthodox economists as “the father of modern finance.” He defended the Efficient Market Hypothesis (EMH), claiming that it is impossible to gain an advantage by predicting stock prices.

The EMH follows automatically from the assumption of rational expectations. If investors are rational, then stock market prices should fully and accurately reflect all available information about the fundamentals of each publicly traded company — and prices should change only when new information enters the market. New information is, by definition, unpredictable (or it would not be new), therefore future price changes must be random.

In 1991, Farmer and colleagues founded the Prediction Company. At that time, they knew very little about Finance or Economics, so they treated the stock market as a vast stream of numbers. Their strategy was to find “predictability patterns” in historical prices using machine learning and other statistical methods. Their performance showed that with the right information and the right model, it is possible to take advantage of hidden patterns in stock prices. They proved that, strictly speaking, the Efficient Market Hypothesis is wrong.

According to Minsky, capitalist economies are inherently unstable due to the cyclical behavior of financial agents. Periods of stability generate behaviors that have the potential to lead to instability. During periods of prosperity, confidence increases, margins of safety shrink, and this leads to rising indebtedness and speculation. The result is financial crises.

To mitigate the inherent instability of financial systems, government intervention is necessary. This includes financial regulation, active fiscal and monetary policies, and a Central Bank acting as the lender of last resort. In a way, as we have already discussed in monetary economics from an econophysics perspective.

To conclude the topic, returning to the discussion about market efficiency, even when using the assumption of rational agents, John Geanakoplos’s Leverage Cycle Theory provides an example of how institutional structure can cause market inefficiency. Even if the market is informationally efficient—that is, if no one can obtain excess profits—it is far from allocative efficiency.

Conclusion¶

Farmer predicts the development of computational policy laboratories capable of integrating data from various sources, including production, consumption, innovation, and financial systems. By leveraging the power of high-performance computing, these laboratories will run complex simulations to explore different policy scenarios and provide guidance for addressing problems such as climate change, inequality, and financial instability.

In essence, Farmer’s guidance calls for a shift away from simplistic and abstract economic models toward a more empirical, dynamic, and evidence-based framework. By embracing complexity and the power of agent-based modeling, he believes that researchers and policymakers can gain deeper insights into how the economy works and develop more effective solutions to the pressing problems society faces.

For this to be possible, one of the main obstacles preventing progress in Complexity Economics is the lack of comprehensive granular data on economic behavior. ABMs require much more detailed information to reach their full predictive potential. Ideally, the models would have access to data such as invoices, receipts, and supply chain information from companies around the world, along with behavioral indicators from households, such as decisions related to home purchases or consumption.

Of course, one obstacle to this is the capitalist tendency to prioritize short-term gains over long-term considerations. There is a need to cultivate a long-term perspective by prioritizing sustainability, social well-being, and the health of the planet. This requires challenging existing norms and incentive structures.

Models, theories, reality and laws¶

I was planning to discuss the paper Econophysics for philosophers in the section Filosofia da Econofísica, but although it is an extremely rich text, I believe that most of its ideas are already well synthesized in the current material. However, there are two smaller discussions that I would like to address, and upon reflection, I came to believe that they fit better in this section.

Models, theories, and reality¶

In his Concepts of Science: A Philosophical Analysis, Achinstein (1968) lists five features that are characteristic of theoretical models in science:

1. It is a set of assumptions about some object or system.

2. These assumptions attribute an inner structure, composition or mechanism, which manifests itself in other properties exhibited by the object or system.

3. These assumptions are treated as a simplified approximation useful for certain purposes.

4. The model is proposed in the framework of some more basic theory or theories.

5. The model may display an analogy between the object or system described and some other object or system.

We are able to connect some of these concepts to the models belonging to the standard approach:

1. Financial markets are the systems described (as modeled by Black–Scholes);

2. The assumptions are understood to be approximations;

3. The model is part of general equilibrium theory;

4. The model sets up an analogy between financial markets and any equilibrium system one could care to think up.

The problem arises at point 2. As McCauley pointed out in the quote below:

No mechanism is held responsible for the properties of financial markets as encoded in the Black–Scholes model.

This is where approaches that depart from orthodoxy can be considered an advance in economic science. This is where the econophysicists’ methodology might be considered an improvement; there is a sense in which many econophysicists are trying to “latch more onto reality.” Their generative models (building economic phenomena from the ground up) are an attempt to understand the causes of the phenomena, and thus we transcend a merely phenomenological description. As in physics, the hope is that by understanding the microscopic behaviour of interacting individuals, the macroscopic structure of complex economic systems can be brought out.

The question of laws¶

As McCauley says, physicists claim that “macroscopic law could arise from total microscopic lawlessness.” There is justified skepticism about this, but it is important to remember that (1) this regularity encoded in a law is usually expected from a collective perspective, and (2) the law one hopes to find has a stochastic nature.

Therefore, the question that arises is whether, at the collective level, the agents as a whole behave in a lawlike way. This isn’t an easy question to answer. Besides which, there is a certain presumption in supposing that there are no laws governing the agents either: Normally, this assumption stems from the belief that people, because they have free will, can do whatever they want, but there will, clearly, be some bounds on what agents are able to do if we look at the situation from a materialist perspective; they will be constrained to behave in certain ways because of the human condition (the social class they belong to, the wealth they possess,etc), the nature of their brains and bodies, and so on. So it is clear that, at this level, there are indeed laws, and these laws will surely constrain the empirical distribution, albeit perhaps in a limited way.

Just as skepticism regarding the existence of laws at the collective level is valid, I believe skepticism regarding their absence is also justified; after all, this too is a supposition about how the world is, presented without much supporting argument. However, econophysics is ripe for examination concerning the issue of laws.