Dodging Economic Reality

How Today’s Economists Conveniently Misunderstand our World

Sean McClure
39 min readSep 25, 2022
Photo by Alexander Grey

You can also listen to the podcast version of this piece on NonTrivial.

Economists are known for attempting to treat economics like a genuine science. But upon closer inspection it becomes obvious that their methods are quite outdated. As a consequence, most of today’s economists are providing an extremely naive “understanding” of our economy, and worse, damaging society’s ability to improve people’s lives.

In this piece, I use Eric Beinhocker’s book The Origin of Wealth, Evolution, Complexity, and the Radical Remaking of Economics, to anchor my conversation around how today’s economists conveniently misunderstand our world.

There are 2 ways to try and understand something. You can attempt to reverse engineer the thing, learning about its components, seeing if you can piece together the causal chains that produce what you observe. The other way is to view the thing as a whole, and notice only its high-level behavior; its attitude, wants and needs.

While the former sounds more “scientific” to most, it’s an approach that has had its time. As I discuss often on NonTrivial such reductionism is as outdated as it is tempting. To reverse engineer a thing is to see it’s components but it is not to see how those components produce the outputs we observe. Reductionism is mathematically convenient, but in all but the simplest systems it does not map to the reality that appears before us.

Almost all relevant (and interesting) phenomena produce their behaviors via emergent mechanisms that do not expose causal paths back to their origins. Particle physics aside, I argue that science should be casting its views towards the whole, not the pieces.

Economics has always had a kind of jealousy of genuine science. Much of its history is loaded with attempts to make it look more rigorous. This effort has brought economics into the world of reductionism, as is the case with all social researchers. And as always, the inevitable outcome of these efforts has plagued the field with misconceptions, bad momentum, and downright untruths.

This is not some passive comment about an unchangeable reality as much as a call for action. Today’s economic models are both deeply flawed and consequential. People’s lives are impacted by really bad models.

The economy is another phenomenon like any other. It contains matter, energy and information, and is thus subject to the laws of thermodynamics. While an individual’s contributions to an economy has little meaning (for most of us) the statistical high-level properties of such a social system are both fascinating and important.

If we want to truly understand the economy we must first accept that we never will in the reverse engineering sense. But we can understand the economy as a holistic complex system with well-defined behaviors; something worthy of modeling appropriately.

Let’s begin.

Greed and Envy

We will begin our economics talk as many others do, with Adam Smith, rightly considered the father of modern(?) economics. Smith was concerned primarily with the creation and allocation of wealth in society. Value creation came about whenever someone could use the resources earth provides to create something people want. The best possible allocation of wealth, according to Smith, comes about when everyone maximizes their self interest.

This is when people like to reference the 1987 film Wall Street and its antihero Gordon Gekko played by Michael Douglas. Gordon’s famous line “Greed is good” is an apparent testament to Adam Smith’s self-interest mechanism.

But framing self-interest in terms of something pejorative like “greed” is shortsighted. Self-interest is not good or bad, it just is. In reality, localized self-interest often leads to better outcomes for the whole. Remember, what the small things are doing is not the same as what the big thing is doing.

Smith’s original ideas come to us via moral philosophy, since understanding how to create and allocate wealth should help us create societies that are the most just for the most people.

But such hand-wavy ideas lack the kind of rigor we see in “true” sciences. After the age of Adam Smith came major advances in the physical sciences, particularly physics and chemistry. This is an age when smart-looking equations and diagrams littered the chalkboards of academia, showcasing explicit interactions, and enabling the kind of bold predictions we now read about in the histories of science.

This was a time when the original insights of thermodynamics began to bear fruit. Tinkerers were discovering new ways to harness combustion, bringing forward a new age of machines and human transportation.

Eventually such trial-and-error discoveries were formulated into theories regarding the behavior of gases. Pressure, volume and temperature could be described using the language of physics, with its force vectors and corresponding equations.

And so as anyone might ponder, if economics were to be taken seriously, as something beyond the moral philosophies of people like Adam Smith, then shouldn’t it be more akin to a genuine science? For something to be taken seriously it needs all those force vectors and fancy equations. It needs to be mathematized.

Along comes Léon Walras and a few others (William Stanley Jevons) who decided they could take what was happening in the physical sciences and apply the same ideas to the field of economics. After all, an economy is a group of individuals interacting in various ways, akin to gas molecules colliding, leading to some interesting and consequential behavior.

I’ve talked about the problem of physics envy before. Physics appears rigorous and “smart” because its chalkboards are loaded with formulae detailing the mechanisms behind the phenomena of interest. But the only reason this happens in physics is because almost all of its phenomena of interest are “simple” systems. Simple systems don’t have causal opacity; one can reverse engineer simple phenomena to see how things “add up.” Even if the phenomena are complex, the physicist’s task is to strip away that complexity and expose the components. Those who choose to deal with components will always get to use more math, because one can express the pieces and their supposed interactions with symbols.

When everyday people cast their gazes upon such formulae they are usually impressed, despite understanding almost nothing about what they see. This has always been the downside of math, though not the fault of math. Math looks rigorous regardless of how well it maps to reality.

And so it was with Léon Walras, who couldn’t resist using the notions of counterbalancing forces to model the economy, applying such ideas to the concept of supply and demand. Since gas molecules interact and settle into equilibrium temperatures so too could people be thought of as agents that drive the market to an equilibrium price and quantity level.

Figure 1 Thinking of the economy like a system of gas molecules was the first (and mainstream) attempt to understand the economy in a more scientific fashion.

At first blush this doesn’t seem so bad. Thermodynamics is fundamental to understanding nature, and not just in the physical sciences. Most real world phenomena are based on collections of agents interacting in complex ways. Thinking of such phenomena as systems means using the statistics of ensembles to predict their outputs. That’s thermodynamics.

But a major consequence of thinking of the economy as a system akin to a gas assumes the economy regularly settles into an equilibrium. Specifically, the “forces” of supply and demand are assumed to counterbalance each other and settle, just as a gas eventually settles into a uniform stable state.

But markets are never truly in equilibrium. Supply never really equals demand. We know this because the economy is actually operated around disequilibrium. Think about it. The economy runs on stocks of inventory, order backlogs, and slack production capacity. There are market makers who continually attempt to smooth the disequilibria that regularly occur. These can all be thought of as delays in an otherwise instantaneous process¹.

But hold on. Despite some deep discrepancies between the borrowed equations of Walras and what we see in real economies, isn’t science about approximation? After all, we all know “all models are wrong.”

The problem is the connection between economic models and reality. I don’t just mean in the sense of how good (or not) the models are, I mean the way policy makers use whatever academics tell them (itself an awful policy).

Traditional economics, rooted in the models of Walras, has had a major impact on public policy, business and finance. Policy makers such as central bankers, presidential advisers and finance ministers regularly rely on these models. In addition, these models are used to inform decisions in the business world related to stockholders as well as competitive strategy. In fact trillions of dollars are traded every day in the markets using calculations that come to us from the “gas molecules” approach of traditional economics. These models are the basis of the interventions we take in attempting to control our economy.

Traditional economic models are fundamentally flawed, for reasons we will get into. But the answer cannot be to not intervene at all, unless of course we all choose to subscribe to anarchist ideology. A complete lack of regulation or government intervention would likely prove problematic. So, is there another option?

Listeners of NonTrivial already know this answer, at least at a high level. It makes little sense to model something that is obviously complex like the economy in terms of basic physical forces. This causes problems in all other areas where physics envy leads to misplaced concreteness.

This came to a head during a meeting in Santa Fe between economists and physicists in the late 80s. The Santa Fe Institute (SFI) is an independent, nonprofit theoretical research institute located in Santa Fe, New Mexico and is dedicated to the multidisciplinary study of the fundamental principles of complex adaptive systems. This includes physical, computational, biological, and social systems. Basically, the SFI was formed to help bring about a focus on complex systems; something severely lacking in mainstream science at the time (still the case).

The meeting between the economists and physicists helped promote the kind of interdisciplinary research that was gaining steam in the late 80s. Disparate intellectual fields could share their ideas on concepts and models, letting others know how they chose to model their phenomena of interest. Cross-pollination of concepts and approaches is always a good idea since solutions to problems never belong to a particular field.

But in this instance, as the physicists and economists shared their methods, something became quite apparent. Economists were using extremely outdated methods to model the economy. You didn’t need to be an economist to know this since their methods were all borrowed from physics to begin with. But the methods they were using were the original ones from Walras, and there had been entire revolutions in science since those methods were used.

One of the physicists famously commented that the field of economics reminded him of Cuba and their outmoded cars; unaware of just how behind they were due to their isolation. In the case of economists it was isolation from the rest of the scientific community.

The ideas of forces leading to equilibria were extremely limited compared to the kind of methods developed since the 2 major revolutions in science; Quantum Mechanics and Relativity.

But it wasn’t just isolation that kept economists tethered to outdated methods. After all, it’s not likely that no economist had heard of the advances in science, some of them likely having degrees in science before focusing on economics themselves. The real reason for remaining fixed on outdated methods was (and is) convenience.

Once you decide your system of interest can be modelled as an equilibrium the math becomes quite basic. The math gets easy because almost all the realistic complexity of the system is artificially removed by the insistence on extreme assumptions.

The assumptions made by traditional economics can be summarized as follows: people will always do what’s in their economic self-interest, and do so in fantastically complex and calculating ways¹.

We will spend some time discussing what this means and more importantly the consequences such assumptions have on economic models. But even a casual glance at such an assumption would make most people wince.

Think about the kind of economic decisions real people make. They purchase products and services, buy homes, etc. Traditional economics assumes that when people are making these purchases they somehow take into account inflation rates, estimates of future government spending, the trade deficit, and so on in their daily decision making.

These kinds of assumptions are the only way one can model an economy as a system that settles into a nice equilibrium. People have to have access to perfect information instantaneously, and make perfectly rational decisions with that information. If they didn’t the system wouldn’t be in equilibrium, since differences in the use of information (e.g. different access times, “irrational” decision making, etc.), would be akin to forces in the system that don’t counterbalance each other.

Traditional economics assumes that there are incredibly smart people in unbelievably simple worlds when in fact it is the precise opposite¹.

Think of the gas molecules metaphor. Instead of perfectly identical balls (molecules) elastically bouncing into each other, imagine balls that are inelastic, with delays in when they bounce and by how much. Some balls don’t bounce at all, while others pickup speed midway through their travels. Such a system would never be at rest. It would exist as some asymmetric collection of particles with a rich variety of behaviors.

By the time the Santa Fe meeting took place it had been 100 years since Walras published his seminal work on general equilibrium theory. What started as physics envy with Walrus was now something even worse, since it was now the physicists who were pointing out how outdated the economist’s models were. It’s like economics was jealous of physic’s old girlfriend who they weren’t even dating anymore. Move on.

The overarching theme here is how traditional economics disregards time. Traditional economics models trade time for mathematical convenience. To understand why this is a problem let’s turn to the age-old joke about an old and young economist:

An old economist and young economist are walking down the street. The younger economist says “hey look a $20 bill!!” while the older (and supposedly wiser) economist doesn’t even look down, and simply says “nonsense, if there had been a $20 bill someone would have picked it up by now.”

This is the idea behind the efficient market hypothesis that says that asset prices reflect all available information. Colloquially this means transactions happen immediately in any financial system. But this is a drastic assumption that anyone should realize strays from reality.

This is because there must be a finite amount of time that passes for any transaction to occur. It takes time for someone to discover the $20 bill, perhaps quite a bit of time. There are likely multiple bills on the ground at any given time, with a wide variation of pickup times.

There is little reason to believe things happen instantaneously in the market. This would mean price discrepancies could be arbitraged away instantly, which is not what we see. Opportunities take time to be discovered, they come and go, they may or may not be worth using, and there are likely a variety of barriers to the transactions.

These barriers to instantaneousness mean there is little reason to expect an economy to be in equilibrium. Yale economist Herbert Scarf actually calculated how long it would take an economy to reach equilibrium and came up with the answer 4.5 quintillion years³. It turns out that the time to equilibrium scales exponentially with the number of products and services in the economy to the power of 4.

In other words the “bouncing balls” with all their differences would take a near infinite amount of time to finally settle into the uniform stable state of the traditional economics worldview.

What is obvious is that traditional economics is founded on extremely convenient mathematics and overly-simplistic assumptions about how economies work. And this is all consequential. The economy is central to how we participate in this world, create value for others, feel fulfilled and define our futures.

When Being Right Most of the Time is Useless

To be fair, the predictions of traditional economics are not totally off the mark. Supply does roughly equal demand and prices do sometimes converge. The markets can act as though they are in a kind of equilibrium.

So what’s the problem?

Why can’t we just use traditional economics as a rough guide to making decisions regarding public policy, business and finance? Why can’t central bankers, presidential advisers, finance ministers and businesses use traditional models to generally inform policies and set business strategies? Doesn’t it make sense to use a model that works most of the time?

The question we have to ask is what happens when the model is wrong? In equilibrium systems the answer is relatively inconsequential. A wrong model means once in a while we miss the mark, but the system can still be expected to return to whatever our model usually predicts.

But in complex systems this is not true. Not true at all. When complexity goes “wrong” relative to a model’s predictions the consequences can be drastic. This is because complex systems have almost all of their outcomes dictated by rare events.

The stock market crash of 1929 took out thousands of investors in one day. The crash on September 29, 2008 was the largest point drop in history prior to covid, wiping out years worth of wealth in one shot. When Terra-Luna collapsed in May 2022 it wiped out almost $45 billion of market capitalization over the course of a single week.

If almost all the wealth gained over years, even decades can be wiped out in a single day then you are not existing in the kind of simple, equilibrium system of traditional economics. You are existing in a system that is extremely “fat tailed”, which means the events that dictate what the market does are both extremely rare and impactful.

This isn’t just markets. Systems in nature exhibit this same behavior because they too are complex systems. Avalanches will accumulate over time until a single event wipes out the mass of snow. Using models that work most of the time is exceedingly dangerous in complex systems if the model doesn’t account for those rare events. Those rare events are everything.

The core problem with models in traditional economics is they only work in well behaved markets/economies. Better, more appropriate models are needed if we are to gain understanding of the economy, and enable policies that actually align with real world complexity. We need the kind of methods used to model complex systems.

The starting point for such an exercise is to show economists that their use of thermodynamics, while itself a sound idea, is lacking a critical ingredient.

I mentioned Walras towards the beginning. He was instrumental in mathematizing the field of economics, and as a result, misapplying basic physics to model a complex phenomenon. But his use of thermodynamics was not a bad idea. Anyone interested in understanding nontrivial systems at a rigorous level should be using a thermodynamic (and information-theoretic) description of that system. As I mentioned previously, social systems are not abstract mathematical constructs, they are real physical systems. Social systems have matter, energy and information, and are thus subject to the laws of thermodynamics.

But Walras only relied on the 1st Law of Thermodynamics (energy is neither created nor destroyed). In his defense, that’s all he had available. But we now know just how critical the 2nd Law of Thermodynamics (entropy in closed systems always increases) is to understanding complex systems.

The 1st law says that if energy is conserved the system is guaranteed to reach equilibrium. Think of a ball rolling inside a wooden bowl. As the ball rolls it dissipates heat, and accomplishes work, giving away its energy. Eventually the ball settles to the bottom of the bowl. The back and forth rolling of a ball until it stops is a good way to think about a system reaching equilibrium.

Figure 2 The concept of reaching equilibrium expressed as a ball rolling down and settling at the bottom of a bowl. Drawing by Sean McClure.

Only if energy is added from the outside (e.g. shaking the bowl) can we kick the system out of equilibrium.

Thinking of the economy in this way means the notion of value is a fixed quantity, that is merely converted from one form into another, just as energy is a fixed quantity that is converted from potential to kinetic energy in our bowl example.

In a traditional economy the earth’s resources are converted into goods, exchanged for money, exchanged back for goods, and consumed. In this scenario new wealth is never created. Instead finite resources are merely reallocated. Recall Smith’s view of morality being anchored on the concept of allocation of wealth.

The above situation is still taught in economics textbooks today. But it’s completely devoid of the extremely important 2nd Law of Thermodynamics, which states that entropy in closed systems always increases. Entropy is a measure of disorder or randomness in a system. If disorder keeps increasing then nothing of value can ever be created since only low-entropy, nonrandom things do useful stuff. Life itself is not possible with pure randomness since large ordered molecules are required for life.

Life exists “despite” the 2nd Law of Thermodynamics because earth is an open system. Open systems allow energy, matter and information to enter into the system, which permits entropy to be lowered locally, at the expense of increased entropy everywhere else (i.e. 2nd Law still holds). Recall my episode on Technology is Humanity where I framed technological innovation in terms of local entropy reduction.

Figure 3 Entropy can be decreased in an open system, since overall the entropy will still increase. Image by Sean McClure.

Order is organization, structure and function, which is opposite the natural tendency towards randomness/chaos (“Schrödinger’s paradox”). The evolution of biological systems occurs in the direction of increased complexity. Survival is the adaptation of species to their environment so as to minimize entropy production.

The creation of large complex molecules, which enables life and all its complexity, happens because the sun continually pumps energy into the system allowing low-entropy things to be created at the cost of increased heat and disorder somewhere else on and beyond earth.

The critical point here is that open systems are not in equilibrium. Imagine hot and cold gas molecules in a container separated by a wall in the middle; hot gas on the left, cold gas on the right. If we remove the wall separating the hot and cold gases we expect them to mix until there is a single, uniform gas at a single temperature. This is equilibrium. This is the most disorder the system can have (the gases prior to mixing were in a more ordered state). The gas example is a closed system.

Figure 4 The relationship between equilibria and entropy. Image by Sean McClure.

If the economy was a closed system its defining characteristic would be a trend towards less order, as with the gas example. We would see less complexity over time. We would expect entropy to move our world from a rich featured environment to a featureless nothingness. If we stopped the inflow of food, oil and information then entropy would be unopposed; our economy would drift towards a kind of equilibrium death. As Eric Beinhocker suggests in The Origin of Wealth, one could argue that countries like North Korea suffer misery and starvation due to their isolation (lack of inflow), whereas vibrant economies like that in the US fair much better (usually), since they exist far from equilibrium (constant inflow).

Existing in a state far from equilibrium is another hallmark of complex systems, and it leads to the kinds of properties we see in our economy. Specifically, systems that exist far from equilibrium exhibit exponential growth, radical collapse and oscillations. These are the signatures of so-called complex adaptive systems and ALSO the distinctive behaviors of our economy.

Resting economic theories solely on the 1st Law of Thermodynamics is mathematically convenient, but it leaves out critical realistic behaviors that must be accounted for. The reality is that the economy is best viewed as an open disequilibrium system, and more specifically a complex adaptive system. Our economy creates novelty (value) as time progresses. Our economy shows signs of self-organization, structure, and increased complexity.

If we are to truly understand the economy we must move away from simple force vectors and equilibria and instead comprehend how complex behaviors emerge from a collection of people who create and reallocate value.

Adding Some Sugar

Joshua Epstein and Robert Axtell are researchers who wanted to see if they could “grow” an economy from scratch. To achieve such a task one relies on computer simulation for obvious reasons. Epstein and Axtell called their pet economy Sugarscape since the thing of value in their fictional world was “sugar”; of the programmed kind. Sugarscape has a physical space in that the agents in this world can move North, South, East and West, and the “terrain” varied by virtue of “mountains” and “valleys”, as well as fertile areas (lots of sugar) and desert areas (little sugar).

The agents in their simulation represent people trading goods. 250 agents were randomly added to Sugarscape, and since there were both desert and fertile areas some were born into sugar wealth and others were not. Each agent was also given a “genetic endowment” for vision and metabolism, such that certain agents could see more steps ahead and utilize the sugar more effectively.

When Epstein and Axtell first let their program run the behavior on the board looked like chaos. Agents randomly bumping into each other collecting and consuming sugar. But eventually order starts to emerge.

The emergence of structure is something we see in complex systems, and Epstein and Axtell’s Sugarscape is a simulation of a complex system. One behavior that emerges is the rich get richer pattern, which I discussed in my episode called Wealth, the Middle Class and the Shape of Networks. In that episode we looked at how the preferential attachment mechanism leads to wealth disparity in real economies. Any simulation that models market dynamics accurately should thus also show a concentration of wealth, which is what Epstein and Axtell saw.

What’s important to realize is that the concentration of wealth in Sugarscape was not due to genetic endowments nor did it matter where the agents started on the board. An agent’s circumstances had no bearing on who found themselves in the top echelon of Sugarscape society. We will touch on this in more detail later.

As I’ve discussed before, the skewed distribution of wealth is an emergent property of complex systems. In the case of Sugarscape it arose from some intricate combination of the environment, the agents, and their interactions. There is no causal story that can be traced back to how such things emerge. They are an invariant, reoccurring property of vibrant economies.

Epstein and Axtell even made further additions to add realism. They added “sex” to Sugarscape by allowing agents to reproduce and pass on their characteristics. This led to the least fit members dying off (running out of sugar), the most fit having more offspring, population swings (cycles of feast of famine) and an even wider gap between the rich and the poor.

So far this was all modelled with the agents as pure hunter-gatherers, collecting and consuming whatever they found on the landscape. But then the researchers added a second commodity, spice. So, in addition to fertile sugar mountains there was now also spice mountains.

They also made it possible for the agents to trade. This was modelled as straightforward bartering, meaning if one agent had a lot of spice and needed sugar and another agent faced the reverse situation then both agents could improve their circumstances by trading, and agreeing on a price. Note that there is no money in Sugarscape, so “price” simply means the relative value of sugar to spice, or vice versa.

Running the program with the above additions initially led to what traditional economists predict. Everyone can trade and overall everyone is better off. More specifically, Epstein and Axtell were able to reproduce supply and demand curves, and the result was the classic textbook downward sloping demand and upward sloping supply.

Keep in mind that the researchers did not explicitly program anything about supply and demand into their model. This is a pattern that emerged purely as a bottom-up phenomenon. It comes about only via agent-agent and agent-environment interactions.

So, to a first approximation most things seem to agree with traditional economics. But closer examination showed that the prices and quantities traded never settled on the traditionally predicted equilibrium point. This would be at the intersection of the supply and demand curves. Instead prices fluctuated in the vicinity of that equilibrium point.

Figure 5 Supply and Demand Curves from Traditional Economics and Sugarscape. Sugarscape reproduced the curves expected from traditional economics (TE), but did not show the idealized equilibrium point predicted by TE. Image recreated by Sean McClure, originally from The Origin of Wealth¹.

Now at this point one might say “well the deviation from the exact equilibrium point is just noise.” But this argument doesn’t hold because there was no noise added to the model. All of the agent’s interactions were perfectly deterministic. Only the initial addition of agents to the board had a random component to it.

So what’s happening? The proper interpretation of these results is that prices move dynamically around an attractor but do not settle into an equilibrium.

This means that the so-called “law” of supply and demand touted by traditional economists is merely a loose approximation. In addition, the so-called “law” of one price breaks down, since Sugarscape showed wide variances in price.

Another tenet of traditional economics is that markets should have so-called “Pareto optimality” where no reallocation of resources can make someone better off without making someone else worse off. But in the Sugarscape simulation it was shown that the market operates at less than Pareto optimality. In other words there always exists trades that could have made agents better off but did not.

If there existed trades that could have made agents better off why were they not executed? The reason is that trades are separated in time and space. So even though trade can “lift all boats” so to speak, making society richer as a whole, it also widens the gap between the rich and poor.

Traditional Economics, and Then Some

As discussed by Eric Beinhocker in The Origin of Wealth new theories should always reproduce the successes of old theories and in addition add new insight; something called the Correspondence Principle. We have seen that models based on complexity, such as Sugarscape, do reproduce many of the elements of traditional economics. Supply and demand worked in an approximate fashion and there were indeed significant societal gains from trade.

But what’s critical is that Sugarscape was able to reproduce these results without the unrealistic assumptions of traditional economics. The agents in Sugarscape were not programmed to have superhuman powers of rationality. There were no preexisting social structures or economic institutions. Critically, it did not assume that everything happens instantaneously.

Sugarscape spontaneously evolved the complex order, structure, and diversity seen in real vibrant economies. It even showed what could be interpreted as “tribes”, “market towns”, “trading routes”, and “capital markets.” Again, none of these were programmed into the system; just simple starting rules and the constitution to let the program run on its own.

The economy is a dynamic system; it changes with time. Prices move up and down, people’s wages change, organizations enter and exit markets. While all these facts sound obvious none of this dynamism is taken into account by traditional models.

The only way traditional economics recognizes the dynamism of real world economies is by treating things like price movements, disruptive innovation, political events, and shifts in consumer taste as exogenous, meaning something that originates outside the system that cannot be predicted/modeled; things like weather events or other “catastrophes” that supposedly come from out of nowhere. Next time you hear a politician, or one of their economists, say “we couldn’t have seen this coming” it’s likely because they are relying on outdated models from traditional economics. After all, most people misuse the term “black swan.”

The dynamism of markets should be expected to emerge from the structure of the economy itself. Whereas the static equilibrium world of traditional economics can only treat dynamic behavior as exogenous, complexity economics shows us that the ups and downs, pulses, collapses and swaying trends of the market are just the type of behavior that usually emerge in complex systems. If your model is to produce a realistic approximation of the economy it should encompass and reproduce its core behaviors, not shuffle them off as inconvenient and uncontrollable variables.

Modeling Complex Behavior

Models of complexity can reproduce price swings, changes in consumer tastes and economic collapses since these are what arise naturally under complexity, although we don’t normally call them by these names. In scientific vernacular we call them positive feedback loops, negative feedback loops, time delays and nonlinearity.

And so to begin modeling the economy properly we must map the parlance of money and markets onto the vernacular of complexity. Economists often think about market mechanisms in terms of so-called “stocks” and “flows.” A stock is anything that can be accumulated (e.g. total money supply, number of people employed) whereas a flow is the rate at which a stock changes (e.g. central bank increasing/decreasing money supply, companies hiring/firing employees).

The stocks and flows of an economy are connected to each other in intricate ways. Imagine employment falling, then a policy maker cutting interest rates to encourage borrowing, increasing the amount of money available for investment, used by businesses to invest in more productive capacity, leading to more demand for employees, raising the stock of employment, and finally affecting future interest rate policy.

Figure 6 The intricate relationships between the stocks and flows of an economy. Image by Sean McClure

Something to notice about this economic scenario is how the output feeds back into the input. We started with cuts to the interest rates from falling employment, which were once again influenced by its own produced effect; increased employment.

This is how feedback systems work. Positive feedback occurs when the situation is reinforcing (hence the word “positive”… not necessarily “good”). A common example of a positive feedback loop is when someone holds a microphone too close to a speaker (the “Larsen effect”). If that isn’t intuitive enough imagine learning to play golf. At the beginning it’s not that enjoyable because you’re not good at it, but the more you play, the better you get, which makes it more enjoyable, which makes you play more, and so on.

Downward economic spirals are caused by positive feedback loops. Imagine a drop in consumer confidence, leading to decreased spending, leading to decreased production, leading to unemployment, leading to even lower consumer confidence, and thus, as per the output-to-input characteristic of feedback loops, a further drop in spending. This pattern can spiral all the way down into a recession.

Positive feedback loops reinforce/accelerate/amplify whatever is happening. Importantly, systems that exhibit positive feedback often show exponential growth, exponential collapse, or oscillations with increasing amplitude. Sound familiar? Markets can turn explosively upward, drastically downward, or fluctuate over long periods of time. This is an example of the kind of behaviors we see in the economy being explained, not as exogenous factors, but as internal (endogenous) dynamics within the system itself.

The opposite of positive feedback is negative feedback, which leads to a dampening cycle instead of a reinforcing one. Think of a system that “pushes” in the opposite direction of some initial direction, bringing things back to a more stable state. A common example is a thermostat, which regulates temperature via negative feedback. Negative feedback loops produce dampening cycles. Think of negative feedback as bringing systems back to equilibrium as they oscillate with decreasing amplitude over time (“peter out”).

Perhaps a more intuitive example; my wife and I just experienced this as we approached a crosswalk the other day. Deciding whether or not to cross comes down to looking at the countdown on the walk sign. If the numbers are low then the decision becomes more questionable (“will we make it”). I thought I noticed her slowing down, which made me slow down, which made her notice me slow down, which made her slow down, and so on. We eventually both stopped and laughed because neither one of us had decided to slow down.

We experienced a dampening cycle where our motion petered out until we stopped. Not because anyone decided to not cross but rather because we got caught in a negative feedback loop. Many processes in biological systems use negative feedback to maintain a desirable state. Examples include homeostatic situations such as thermoregulation, blood sugar regulation and osmoregulation.

Dynamic systems also have time delays. Eric Beinhocker in The Origin of Wealth uses the challenge of finding the right shower temperature as an example. Think about how one overshoots the cold and hot settings, back and forth, until they finally get the water temperature comfortable. We struggle with this because of the delay in response between the water temperature and our actions. Obviously the longer the delay the more difficult it will be to control the shower temperature.

The economy is filled with delays. Recall the joke about the $20 bill; it takes time for someone to discover the $20 bill. The drastic assumptions in traditional economics are founded on the notion that transactions happen immediately in any financial system. But there will always be a finite amount of time that passes for any transaction to occur. Remember, the economy is in disequilibrium, not equilibrium. As stated earlier, the economy runs on stocks of inventory, order backlogs, and slack production capacity. These are delays in an otherwise continuous process.

A critical thing to realize is that these delays are required for an economy to run smoothly. Buffering of stocks (e.g. order backlogs) allows for more continuous flows of the economy at the aggregate level. This is akin to how buffering works in a streaming service like Netflix. In order to ensure a good user experience there must be constant video play even when there is a momentary drop in the internet connection. This is only possible with a backlog of preloaded content. Instantaneousness kills many otherwise effective systems.

The final behavior listed previously in our mapping of economic parlance onto complexity vernacular is nonlinearity. The economy is undoubtedly a nonlinear system. A nonlinear system is a system where a little change in the input can have dramatic changes in the output, and vice versa.

Complex systems are not just nonlinear, since even static systems can produce curved behavior over time. Complex systems like the economy are more appropriately modelled as nonlinear dynamic systems. Nonlinear dynamic systems produce a wide variety of behaviors (modes).

To get a sense of the various behaviors that nonlinear dynamic systems produce we can look at the quintessential example of chaotic systems, the double pendulum. These systems swing out in mostly unpredictable patterns.

Figure 7 The double pendulum as a basic example of a nonlinear dynamic system.

While the economy has chaotic aspects to it, it is not a fully chaotic system. There are far more degrees of freedom in an economy than there are in a double pendulum. The degrees of freedom are the number of independently variable factors that affect the range of states a system can exist in. The economy is loaded with a massive number of interacting factors that come together to produce emergent behaviors.

Systems with a staggering number of degrees of freedom exhibit complex modalities. This means if the parameters of the system are tweaked even slightly we see dramatic changes in output. Imagine drawing out the lines on a graph as we typically do with the double pendulum, except this time for genuinely complex systems. The behaviors we would see would be lines that rise then settle (fixed-point attractor), lines that produce regular oscillations like a pendulum (a periodic limit cycle), lines that show oscillations within oscillations like a heartbeat (a quasi-periodic limit cycle), and lines that appear random for a long time then eventually repeat themselves. And we would see some instances of complete chaos (deterministic, never repeat, but still bounded).

The economy has multiple “stocks” and “flows”, interacting in intricate ways, showcasing positive and negative feedback loops, time delays and nonlinear dynamics. The economy is a complex dynamical system, not a basic double pendulum, and most definitely not a simple equilibrium system.

This all amounts to a core truth about nonlinear dynamic systems: they have extreme sensitivity to initial conditions. Nonlinearities cause small differences in initial conditions to be magnified dramatically over time. Starting a double pendulum from 2 slightly different starting positions produces swing patterns that are wildly different from one another. Real-world economies, as complex systems, would manifest these path discrepancies to the extreme.

This is why the economy cannot be modelled analytically. It must be treated appropriately using computing approaches similar to Sugarscape. In other words, there are no shortcuts when it comes to modeling the economy (like equilibrium theories). You have to run the program out, using the computer’s ability to attempt a massive number of configurations and witnessing what emerges.

It’s important to realize that almost all systems in nature are nonlinear and dynamic. Despite the copious use (and promotion) of linear models in the sciences genuinely linear systems are exceedingly rare. Anytime someone starts promoting a model for something even remotely nontrivial, look for linearity and Gaussian distributions; if you seem them, run away and warn others what you saw.

As Eric mentions in his book¹, the mathematician Ian Stewart thinks having a domain in physics called “nonlinear systems” is silly. It’s like biology having a field called “the study of nonelephants.” Almost everything of consequence is nonlinear and dynamic.

This is why computation is a MUST tool (if not THE tool) for doing science today. Only computation can allow us to see how nonlinear dynamic systems evolve.

It’s not that traditional economics hasn’t recognized the existence of nonlinearities, but they have struggled to incorporate them in any meaningful (dynamic) way. Their recourse is to either use nonlinear relationships inside static models, or use linear relationships inside dynamic models. Both make the solving of the equation fairly straightforward, but there is little reason to believe they map to reality.

Asymmetric Equality

The models of traditional economics should not be used to teach, advise, or intervene on our economy. They certainly shouldn’t be used to attempt to comprehend our economy. They are outdated, dangerously naive and propped up by the illusion of control.

The economy is obviously a complex system, and as such exhibits the properties we know occur in complex systems. The overarching lesson when it comes to complexity is always the same; it isn’t about control it’s about acceptance. We must accept that nature operates the way it does, and look to work with nature rather than against it.

In an economic setting this means admitting that we live in a world that does not provide many levers to manipulate the outcomes of events. The economy is not the cogs and pistons world of the industrial revolution, it is more akin to an ecosystem that presents stressors we must adapt to. Adaptation does not happen by finding root causes and applying specific changes to control outcomes. It works by embracing variation, selection and replication in an effort to produce something that survives. It works by a high-level process that doesn’t naively reach into the internals of a system.

Of all the properties we looked at, exponential growth, radical collapse, oscillations, positive and negative feedback loops, time delays, nonlinearity and fat tailed events, what they all have in common is they emerge. If we setup computational experiments like Sugarscape we can reproduce economic behaviors without programming them into the experiment. They ALL arise from basic local decisions of the agents involved.

This means that whatever we see in our economy, whether it’s wealth disparity, disruptive innovation, stock market collapses, etc., these are all inevitable behaviors that exist because complex systems produce these kinds of outputs to survive. Recall my episode on Things Only Look Crazy When You Stand Too Close where I discussed the “cost of complexity.” The mechanisms that make complex environments tractable are not free; they come with events and behaviors that humans deem “bad.”

We must accept these behaviors as being there for a reason, without knowing the reason. We must resist the scientism that plagues today’s elitist and naive view that we can control nature.

Take wealth inequality. This is a perfect example of the post-modern belief that we can intervene in an otherwise natural process and create better outcomes for everyone. This belief is founded on the notion that there are levers we can pull to outsmart nature, as though social engineering hasn’t shown us enough disasters. Rest assured, forcing equal outcome is fully unscientific and guaranteed to fragilize any system in the long run. This is baked right into the probabilistic foundations of how networks function.

Wealth inequality is a fully inevitable outcome of complexity, reflecting the Pareto asymmetries we see in complex networks. Recall my episode on Wealth, the Middle Class and the Shape of Networks where we looked at these asymmetries as they relate to enterprises and their labor.

Figure 8 Wealth disparity is reflected in the distribution of people who do and don’t have “wealth.” The left image is a depiction of a network, with circle size representing wealth. The right image shows a corresponding conceptual plot of the wealth distribution.
Figure 9 The economic disparities apparent in real economies show wealth concentrated in the “peak” of the lopsided wealth distribution.

But this doesn’t mean there are no solutions to wealth inequality. Complexity doesn’t mean throwing your hands up in epistemic resignation admitting defeat to nature, resigning oneself to whatever station in life one was assigned.

Recall the Sugarscape experiment, where agents roamed around a grid looking to improve their situation by acquiring “sugar.” Some agents were born close to the “sugar mountains” and some were given “genetic” advantages such as better “vision” and/or increased “virility.” Most people might assume the results of such an experiment would produce 1 of 2 outcomes:

  1. that the random movement of agents would result in everyone equally finding enough sugar to prosper; or,
  2. a wealth disparity would appear from the advantages given to those born close to the mountains and/or having genetic advantages.

But neither of these were the result.

There was indeed a wealth disparity, which as I just discussed should be fully expected. BUT, those who benefited the most came from “all walks of life.” If didn’t matter where they started on the board or what supposed advantages they were given. In the end it was a random mix of agents in the peak of the wealth disparity distribution.

What Sugarscape and other models of complexity show us is that while the asymmetries in the economy are a natural and inevitable outgrowth of complex systems, this does not mean only certain individuals must succeed. Accepting wealth inequality is not the same as accepting that only certain people will be prosperous.

This is because natural networks are not static things, they are dynamic. The nodes that attract the most connections are not the same agent over the long run. Different agents come in-and-out of the wealth distribution’s peak, but the peak remains. It’s not the people in society that are invariant, it’s the asymmetry that’s invariant.

Models of complexity have a way of showing us how complexity is supposed to behave. This is because the models used in complexity don’t rely on explicit programming beyond basic scaffolding. This means there is much less chance for naive intervention to artificially control how systems evolve. Beyond a few high-level constraints and assumptions models of complexity are “let loose” allowing the computer to run millions of iterations and converge on a result over time.

As long as the local interactions of the agents are kept basic, and the computer allowed to run many iterations, the results of computer simulations should be considered a much closer approximation to how nature evolves over time, compared to some analytical model with predestined outputs designed by people. The behaviors that emerged in Sugarscape, like different agents prospering at different times, are thus arguably the kind of properties we should expect in a healthy economy.

But long-term income statistics show this isn’t the case, especially in the United States, the biggest economy in the world (currently). Whereas models like Sugarscape show different agents coming in-and-out the wealth peak, the US is much “stickier.” Social mobility appears quite restricted in the US.

Studies show that mobility opportunities are different for poor and wealthy children. Parental incomes and their choices of home locations while rearing children are major factors in wealth disparity. A 2012 Pew Economic Mobility Project study found that 43% of children born into the bottom quintile (bottom 20%) remain in that bottom quintile as adults. Similarly, 40% of children raised in the top quintile (top 20%) will remain there as adults.⁴

The so-called American Dream is a strong narrative in the US. Apparently only 32% of Americans agree with the statement that forces beyond their personal control determine their success.⁵ But the American Dream is just that, a dream; one that is almost guaranteed (statistically) to not become reality for an American citizen.

Social mobility is largely anchored on educational opportunities certain individuals are granted⁶. With a degree in hand these individuals can seek out better jobs, make better money, and place themselves into a higher social class. Education is undeniably a leading factor in one’s ability to move up the ladder in society.

And therein lies the problem. In my opinion it’s the grossly exaggerated importance society (especially Americans) place on institutional intelligence that is largely to blame for the lack of social mobility. Americans have equated one’s education to their intelligence. They have created an entire economy that uses higher education as the gateway to opportunity. The more schooling you have, the “smarter” you are. The better the school you went to, the “smarter” you are.

Once you institutionalize intelligence you do 2 things:

  1. you make it so one’s current economic standing (often via their parents) dictates the ability to enter the economy (only some can afford education); and,
  2. you ensure that only certain people who excel at a very narrow definition of “smart” are given the opportunities.

This is what I call “bad momentum.” By encouraging this narrative around institutional intelligence we artificially intervene in an otherwise natural process. We prop-up patterns that are not as nature intended. This will always, in the long run, produce bad outcomes.

But isn’t a natural outgrowth of complex systems the rich get richer mechanism? Isn’t the preferential attachment of education and opportunity to those with existing money and a certain type of “smarts” a reflection of this inescapable reality?

But this confuses informational asymmetry with physical asymmetry. Let me explain.

Think of a wave, say a big one from a tsunami. The peak of the wave is maintained for a long time, but the water molecules that form the peak are constantly changing. The peak of a proper (natural) wealth distribution is not a static thing; it is not formed via the physical allotment of certain individuals. The peak just means that the probability of finding someone with a lot of wealth is only high for a few individuals at any given time.

If you could label each water molecule that makes up a wave you would see constant mobility that maintains the peak. The dynamic nature of complex systems means agents do not stick around; they swarm in-and-out of locations to produce informational invariance (“we call that thing a wave”), not physical invariance (“what’s it made of?”). In other words, the static look of a wave is more of an illusion caused by the constant motion of different water molecules swirling in and out of the spot.

THIS is the asymmetry that occurs in complex systems. Not physical people and organizations but concentration of wealth itself, whose form is defined by highly mobile and varying individuals.

Figure 10 Social mobility should be thought of in terms of movement (or lack thereof) in and out of the peak of a fat tailed distribution.

Institutionalized intelligence is both expensive and has a deeply flawed notion of what constitutes “smart.” This bakes-in bad momentum because money must be available to would-be students at the time they get accepted into institutions. This makes a student’s opportunity fully dependent on a family’s history, not on their own ability to generate wealth (since they are too young to do this). Scholarships don’t help since their qualifier is high academic achievement, which gets us back to filtering on individuals who already conform to a restricted, low-dimensional definition of “intelligent.”

The economy cannot function as it should under the narrative of institutionalized intelligence. Such contrived gatekeeping to opportunity artificially restricts the mobility that would otherwise be available to agents inside a naturally occurring complex system.

The competition and selection of natural systems do not function via artificial constructs like educational institutions, they function by competing and cooperating for the creation of genuine wealth. Nature does not need some false narrative to filter an agent’s movement into the possibility space of our economy, and doing so will cripple the adaptive ability of our economy in the long run.

One might attempt to use my argument to support their claims that equity policies like affirmative action are therefore good. After all, the “bad momentum” I speak of is a kind of systemic obstacle to a properly functioning system. Why not remove that bad momentum by taking it away from the privileged?

But establishing better representation inside the peak of a wealth distribution would do nothing to improve mobility; it would only momentarily change who gets to be in the peak. This is still a naive intervention, and would fragilize the system in the long run. We know this, because we know how complex systems evolve and we know that the economy is a complex system.

Mobility means mobility, it doesn’t mean changing who gets to sit in the king’s chair. Even perfect representation has nothing to do with mobility. If one family from every culture sits at the peak, it will be those families who pass on their privilege, with almost everyone in the population, regardless of culture, left out.

Giving the currently underrepresented among us better access to academic opportunities cannot solve the problem, because it does not solve the core issue; institutionalized intelligence. It is the equating of education to capability that must be destroyed.

I’ve talked about the pseudoscience of IQ elsewhere. To be clear, my take has absolutely nothing to do with race, and everything to do with the statistical and scientific flaws baked into such studies. Equating human intelligence to some standardized test score is wholly circular; performing better within the confines of institutionalized intelligence guarantees you will be given better opportunities in life. There is little mystery around the “predictive” power of IQ and job success when our modern economy only grants opportunities to extremely scholastic individuals.

Good representation will arise naturally in complex systems because variation is a core ingredient in how nature solves problems.

A Final Thought

The models of traditional economics, with their convenient reliance on ideas like equilibrium and force vectors, keep society bound to the illusion of control. They encourage social engineering because they tell us there are deterministic levers we can pull to control outcomes. Anything beyond what these models predict, which is much of the economy, are deemed exogenous and uncontrollable.

We use simplistic models as gatekeepers to opportunity because they creep into how we filter individuals in society. This is not some progressive opinion, this is factually how complex systems work.

Economists have a massive influence on how government and businesses function. A lack of social mobility is but one example of the kinds of problems caused by outdated thinking. We cannot tout the virtues of meritocracy when our opportunities are fully dependent on some outdated credentialed form of perceived competence.

So what can we do? We can embrace realistic, but not always convenient, models of complexity. We can appreciate how nature actually works, and form our policies and actions around this understanding.

One aspect we would have to accept is the intermittent form of an individual’s wealth and opportunity. As with water molecules moving in-and-out to form a wave, people should only be expected to run into wealth for short periods of time.

This is very different than how our economy thinks of economic stability and health. We seek jobs that pay good ongoing salaries. We are given lines of credit and mortgages IF we can demonstrate continuing income and payments on debt over time. At the individual level we are expected to land an opportunity which provides us enduring and growing economic status.

But added to the list of complex behaviors I already mentioned, intermittency is a fundamental modality. Nature tends to produce its behaviors in fitful, infrequent and occasional bursts. There is a strong argument to suggest that it is more natural for people to run into their economic opportunities once in a while, rather than land a good job maintained until retirement.

This is inline with one of the consequences of fat tailed distributions I discussed above. Complex systems have almost all of their outcomes dictated by rare events. Having an economy where an individual’s wealth is largely determined by a few instances of economic success may be the most natural economy of all.

After all, do we really need to make good money continuously? Or do we just need a few windfalls to bring about prosperity in our lives? Does the same company need to be around for more than 10–15 years?

For the individual this suggests entrepreneurism is a better path than traditional employment. For organizations it suggests companies shouldn’t be allowed to become artificially too big for too long. Practices like VC funding and lobbying may be other sources of stifled mobility.

Are we destined to all be self-made, intermittent heroes of our own economic journeys? Will anti-trust laws make a comeback to keep competition more natural? Or will the illusion of stability from misplaced models keep us bound to naive notions of wealth, at the expense of almost everyone in society?

Will the false levers of traditional economics prove too tempting for society to even try?

Time will tell.

Suggested Reading

  1. The Origin of Wealth: The Radical Remaking of Economics and What it Means for Business and Society by Eric D. Beinhocker
  2. The Santa Fe Institute
  3. The Computation of Economic Equilibria
  4. Pursuing the American Dream: Economic Mobility Across Generations
  5. Socioeconomic mobility in the United States
  6. Social mobility
  7. Americans overestimate social class mobility



Sean McClure

Independent Scholar; Author of Discovered, Not Designed; Ph.D. Computational Chem; Builder of things; I study and write about science, philosophy, complexity.