Complex Economic Evolution
(Excerpt from my forthcoming book “An Economic Journey into the Future”)
The economy as a complex system worldview has interesting implications for the old debate about the intervention of the government in the economy or not. It presents arguments against and in favor of interventionist and liberal economic policies.
Let us first look at the arguments in favor of liberal policies.
Neoclassical theory argues in favor of non-interventionism and free markets based on the General Equilibrium Model. Should it not have much intervention from the government, the market process itself tends to bring the economy to an optimal situation. In this context, any intervention by the government would only come in the way of hindering this process. The proof for this is done in deductive, rigorous and mathematical form.
In the complex view there is no guarantee that the market process takes the economy to an optimal solution. There is no logical way to conclude that the best policy is non-interventionist. Arguments in favor of liberal policies from a complex perspective are two fold, that of ignorance and history. The argument of ignorance is as follows: the economy is a Complex Adaptive System and is above our formal analytical capabilities, it emerges from the complex interaction of independent agents. It would be a delusion to find that we can affect something as complex as the market in a positive way. One other aspect of this argument would be regarding the concept of “Optimal”, the goal which we should be aiming for. The complex approach leads us to see the economy as a dynamic process that self-organizes, where fluctuations cannot be avoided because they are part of the system. The best state for a capitalist economy is the one at the edge of chaos, where creativity and chance coexist. The edge of chaos is the point where the economic system is stable and flexible at the same time. Fluctuations of all sizes and duration occur in a decentralized manner and are absorbed by the structure of the system.
The historical argument is that whenever agents tried to interfere with the market, it ended up creating more problems than solutions. Just look at the subprime crisis in 2007 caused by the higher demand for housing given the very low interest rates policy the Federal Reserve prescribed to the economy to rescue us from the burst of the internet bubble in 2000. A classic Butterfly Effect. Using history as a guide, the best solution would be non-interference. From the complex point of view, one wants to advocate for liberal non-interventionist policies through the argument of ignorance and history, and not via logically deductive theories.
Now let us look at the arguments against liberal policies.
These arguments arise from the impact of factors such as sensitivity to initial conditions, increasing returns, market imprisonment and path dependent trajectories. The mere existence of these factors questions the idea that the market acts fairly. By and large, people believe that in a market economy each individual earns what it deserves, thanks to its competence and efficiency in a meritocratic regime. This would provide the foundations for the popular support for market economies. The Neoclassical view would be the theoretical proof of the advantages of this regime. From the complex point of view, the results coming from the distribution of wealth derived from the market mechanism are not necessarily the fairest ones. Luck, that is, being in the right place at the right time and being the first to enter a certain market can be the main reason for wealth accumulation. Restrictive practices can also have influences in this process making efficiency and competence secondary factors. There could be the edge case where in a group of 100 equally efficient individuals, one gets all the wealth, and the rest are left with nothing.
A second argument in favor of interventionist practices would come from the effect of increasing returns. In Neoclassical theory, given the presence of decreasing marginal returns, there is a natural limit to monopolies. The competition within a given sector of the economy would lead to an equilibrium. This is not the case in the complex worldview. Free competition can lead to monopoly in certain industries, as is the case with Facebook and Google, as a result of path dependent trajectories in the economy. An economy could follow an unwanted path and a redirection to the best path could be done right at the beginning of this process. The examples of the QWERTY computer keyboard and the Beta vs VHS video standards show how historical accidents can cause future damage to society. The new vision offers the possibility of preventing situations like these from perpetuating themselves and can be avoided through policies that guide the economy into more favorable trajectories.
Complex models based on the interaction between agents have been proposed to help guide government policies on issues such as poverty and social inequality. Steven Durlauf, an economist at the Santa Fe Institute created a computer program that simulates a set of individuals and how these interact, considering characteristics of these individuals such as education, race, familiarity, neighborhood, place of work and so on. Durlauf was seeking the statistical regularities or patterns of the aggregate behavior of their “agents” in the process of forming communities. Just like water molecules interact and in certain conditions form ice, social agents interact and in certain conditions create social inequalities. Both would be emergent properties of the systems. The basic idea of the model was summarized like this[1] :
“ First, individual preferences, beliefs and opportunities are strongly influenced by the participation of an individual in various groups. Such groups can be fixed, such as their race or determined by the economy or society, such as religion, neighborhood, schools or companies. Second, the interaction between agents of a same group influences the economic performance of these agents. Third, social stratification by income, race, education, or language leads to divergence in the characteristics of groups, which results in increased inequalities and decreases social mobility.”
Because Computational Economics accepted for individual utility functions that were different for each individual, agents based modeling was able to simulate emergent behavior in Economics.
Looking at communities under these optics, the opportunities that are passed down from generation to generation can present feedback effect. Given the feedback in the distribution of income and profession that members of a community pass on to their children, there is a great incentive for these to want to live in the most affluent communities. Therefore, big incentives exist for economic segregation in communities. One consequence of this theory is that entire generations of families can fall into poverty and so remain indefinitely, while others will always be rich. Since rich and poor families are separated by the price differential of housing, they become isolated. This isolation means that the children of wealthy families will have better schooling and job opportunities. In the age of AI, they may even have a micro chip installed in their brain to give them superpowers. Therefore, early differences in adults income can be magnified for their children. Ghettos can emerge and perpetuate because of this stratification. We then noticed the presence of path dependent trajectories with the presence of lock-ins, characteristics of a complex dynamic process attracted to a sub-optimal solution for most. In fact, the economy is so complex that trying to modulate it in all its nuances on a computer is a very difficult task but not impossible. Even if Durlauf achieves his goal, we know that small interferences in the economy can cause historical displacements (Bubbles and Crashes) that can affect the long-term predictive accuracy of such models. That said, these kind of models can assist in the arduous task of understanding the consequences of government policies. Current (2020) AI agent based simulations of the economy utilizing Deep Reinforcement Learning techniques done by economists from Harvard and Salesforce were published in a paper named “The AI Economist: Improving Equality and Productivity with AI-Driven Tax Policies”. In the paper they created a simulated economic environment and trained artificial social planners to discover tax policies in dynamic economies that can effectively trade-off economic equality and productivity. In the simulations, both agents and government learn and adapt. The interesting thing about the paper is to show how agents based modeling can be used to simulate an economic environment and inform monetary and fiscal policy in the future. We will go into more detail about the economics of AI in the next chapter.
Just as Darwin’s Evolution Theory does not predict the formation of species but helps to understand the diversity of nature, simulations of the economy can help elucidate it without necessarily possessing long term predictive power in complex environments. However, they are the best proxies for a real lab where experiments can be done. According to the economists, “the simulation framework can also be used to speed up experiments with existing proposals for tax systems, validating assumptions and offering the ability to test ideas that come from economic theory.
Another Santa Fe Institute and Oxford University professor that is one of the founders of the study of Chaos and Complexity, J. Doyne Farmer, is currently working on creating a bottoms up agent based simulation for the whole economy. If anyone is qualified to tackle this immense task, that would him. Farmer is notorious for his work in beating the Casino on roulette game using Chaos Theory and later for beating the markets at Prediction Company, a Quantitative Hedge Fund he founded after 10 years at Los Alamos. I had the opportunity to work with some of his ex-co-workers when I was a quantitative portfolio manager at Millennium Management which later bought the Prediction Company. Over there, we were able to harness the power of Natural Language Understanding (NLU). An AI type algorithm that was programed to read and interpret the sentiment of public companies earnings announcements in real time and decide whether or not to make an investment in the shares of those companies without human intervention.
In the next chapter, we will embark on a journey into the Age of AI, where eventually (I believe in less than 1 year) there will be a digital twin of the Global Economy that could be used as a digital Lab for simulations and predictions. But for now, let’s set the stage and hear from complexity theory economist W. Brian Arthur himself:
In a brilliant speech for the Stanford Complexity Group in March 2016 entitled “Complexity and the Economy”, Arthur summarizes the new economic paradigm that has become the intellectual framework for the economics of the Intelligence Revolution:
“Complex economics features an economy that is not necessarily in equilibrium. In fact it’s usually in non equilibrium. Agents are not all knowing and perfectly rational. They must make sense of the situations and explore strategies as they do so. The economy is not a simple container of its technologies. It forms from them and changes in structure as it happens . In this way the economy is organic, one layer forms on top of the previous one. It’s ever changing. It shows perpetual novelty. It’s attracted to the Edge of Chaos.
Attractors within it appear, persist for awhile and melt back into it again. All this is not just a more poetic humanistic view of the economy. It can be rigorously defined and precisely probed and analyzed. This approach does not fit the standard economics. It’s a different approach and it’s not just the use of computers to do agent based modeling nor of adding a deeper understanding of technology change to endogenous growth models. It is economics done differently. Economics based on different concerns, particularly on how non equilibrium works on economics where the problems are different and the very idea of a solution is also different. One way to see this is to recognize that standard Neoclassical economics comes out of a particular way of looking at the world. Neoclassical economics inherited the enlightened view that behind the seeming disorder of the world lay order and reason and perfection. And it inherited much from the physics of the late 1800s in particularly the idea that large numbers of interacting identical elements could be analyzed collectively via simple mathematical equations. By the mid 1900s this led in turn to a hope that the core of economic theory could be captured in simple mathematically expressed principles and thereby axiomatized. Some parts such as macroeconomics or the theory of institutions might have to be left out for the core of the field could be ordered and tamed and reduced to mathematics.
The program was at best only partially successful. Is certainly cleaned up much of the sloppy logic that had passed as theory before and led to a fresh respect for the workings of markets and for the inherent advantages of the capitalist system. But it also led to stiffness in thinking to a righteousness in what was permitted as economic theory and was not. Also led to a closeness of other ideas, shut were the effects of the economy of politics, of power, of class, of society, or fundamental uncertainty and the formation and creation and development. In the end it could be argued that the program, at least the extreme hyper rational vision of it, failed. Nobody could have claimed in 2008 that the market had lost half of its worth, in a short time, because companies had suddenly lost half of their usefulness. Companies were worth just as much as before. Nobody could claim either that unemployment rates of 20% and upward in some of the European economies were due to sudden changes in preferences of the labor force. People wanted a job just as much as before.
The complexity perspective replaces agents in well defined problems with well defined probabilistic outcomes using perfect deductive reasoning and thereby arriving at an equilibrium. We now have agents who must make sense of the situation they face, who need to explore choices using whatever reasoning is at hand, and who live with and must adjust to an outcome that their very adjustments may cause perpetually to change. They co create the reality they live in.”
Their thoughts and actions create their reality…
One of the earlier insights of economics that goes back to Adam Smith is that aggregate patterns form individual behavior, an individual behavior in turn responds to this aggregate patterns. There’s a recursive loop. It is this recursive loop that connects with Complexity. Complexity is not a theory but a movement in the Sciences that studies how the interacting elements in a system create overall patterns and how overall patterns in turn caused the interacting elements to change or adapt. Complexity asks how individual behaviors might react to the pattern they together create and how the pattern will alter itself as a result of this. Is often a difficult question we’re asking: How a process is created from the proposed actions of multiple agents? And so economics earlier in its history took a simpler approach. One more amenable to mathematical analysis. It asks not how agents behaviors would react to the aggregate pattern they created but instead, what behaviors, actions, strategies, or expectations would be consistent with the aggregate patterns that they caused. In other words, what patterns would call for no changes in micro behavior and would therefore be instead stasis or equilibrium? General equilibrium theory asks what prices and quantities of goods produced and consumed would be consistent with or would pose no incentives for change to the overall pattern of prices and quantities in the economic markets. Classical Game Theory asked the best course of action for an agent given the strategic moves or locations his rivals might choose. And Rational Expectations economics ask what expectations would be consistent with or would on average be validated by the outcomes this expectations together created. The underlying assumptions were often chosen for analytical convenience. Under equilibrium, by definition, there is no scope for improvement or further adjustment. No scope for exploration, no scope for creation, no scope for transitory phenomena. So anything in the economy that takes adjustment, adaptation, innovation, structure, change, history itself, must be bypassed or dropped from theory. The result may be a beautiful structure but it is one that lacks authenticity, aliveness, and creation. What if economics allowed the wider possibility and asked how agents in the economy might react to the patterns they together create? Would this make a difference? What would we see then? We would see endogenously generated non equilibrium for two reasons: one is fundamental uncertainty and the other is technological innovation. The mental uncertainty caused by reflexivity engenders further uncertainty. As Keynes said “we simply don’t know”. And Schumpeter in 1912 famously pointed out that there is a source of energy within the economic system which would disrupt any equilibrium. That might be a technology or combination of technologies. Important to notice the self reinforcing nature of this process. Technology disruption is self reinforcing. Examples of self reinforcing phenomena are bubbles and crashes. Another one is clustered volatility in equity markets. Periods of high volatility followed by periods of high volatility. Periods of low volatility followed periods of low volatility. Until the system reaches a critical point and phase transitions occur. A third one is sudden percolation which means the more connected a network is, the faster change propagates. This leads to a general property of complex systems which is critical levels in phase transitions at the edge of chaos. Another property is positive feedbacks or increasing returns on scale that can cause lock ins, like psychological herd effects caused by fear and greed that culminates in bubbles and crashes. For Keynes, positive feedbacks could lock in the economy into 2 states: full employment or unemployment. The Gods and the Useless, as History Prof. Yuval Harari from the University of Jerusalem describes in his dystopian vision of the future caused by technological lock in where the owners of AI will have superpowers and the 99% will be out of jobs or performing underpaid jobs due to the automation of cognitive intelligence previously supplied by human labor.
On a more positive note, Jeremy Rifkin, author of a variety of great books on the subject of technological disruption, including the 2011 best seller and highly influential “Third Industrial Revolution”, three technologies have to emerge at the same time and converge into a general purpose platform that fundamentally changes the way society manages, powers and moves its economic, social and governance life.
Jeremy sees the economy as a living organism, like J Doyne Farmer. For Farmer, the economy is the metabolism of civilization. For Rifkin, our infrastructures are really a social incarnation (or a fractal) of what every cell in an organism must do, but on a bigger level. As within, so without. As above, so below.
Every cell of every living creature must have:
1. A way to communicate
2. Have a certain amount of energy to maintain its non equilibrium state
3. Have mobility in order to function
4. Have a semi permeable membrane, its skin, so that it can orchestrate its inners with its outers and be attuned to the circadian rhythms which guide the earth.
Get Ilan Gleiser’s stories in your inbox
Join Medium for free to get updates from this writer.
That’s what infrastructures do. It’s not just about building roads and bridges.
The purpose of infrastructures is to bring larger collectives of our species together so we can communicate over time and space and govern our economy and society.
In order for an economic paradigm shift to happen, the three technologies that must emerge at the same time are:
#1 new sources of energy — to more effectively bring local activities together as a social unit with the power to manage the economy
#2 new motion mobility and logistics — to more effectively bring all activities together so that we can move commerce and social life in larger units.
#3 new communication networks — to connect it all together.
When new communication networks converge with new energy sources and new modes of mobility and logistics, it changes the way society and civilizations manage, powers and move its day to day life.
For example, let’s look at the First Industrial Revolution in England:
Communications and Mobility allowed us to go from small villages to large collectivities. Small markets gave rise to nation states to secure those markets.
The three technologies that converged were:
#1 Energy — Steam engine: that enabled the power printing, which gave rise to the public schools and was powered by coal
#2 Mobility and Logistics — Railroads — The steam engine on tracks and also powered by coal
#3 Communication Technology- The Telegraph system
The convergence of these three technologies changed our thin membranes and made us move from small rural areas to large urban cities because of the hub rail traffic. This revolution changed our temporal spatial orientation.
As the economy evolved, we saw the Second Industrial Revolution in the United States in the 20th century. We moved from rail roads to interstate roads, shopping malls and suburban life. This model peaked when oil reached $147 a barrel in August 2008. That was the economic earthquake of the industrial era of the fossil fuel system. The financial collapse, known as the sub prime mortgage collapse, was an aftershock. We could not keep the fictional economy going when the real economy turned down.
The three general purpose technologies that converged to give rise to the emergence of the Second Industrial Revolution were:
#1 Energy — Cheap Texas Oil
#2 Mobility and Logistics — Interstate Roads which Henry Ford populated with combustion engine cars for transport and logistics
#3 Communication Technology — Telephone, Radio and TV
Fossil fuels and nuclear power, Internal Combustion road-rail-water and air transport and telecommunications, formed the infrastructure that manages, powers and moves most of the economy, society and governance until now. This paradigm peaked in its Aggregate Efficiency in every industrial country in the world around between 1995–2005. By Aggregate Efficiency, Rifkin means:
“From a thermodynamic point of view, we are continuously borrowing low entropy inputs from the planet — that can be a rare earth, a metallic ore, or a fossil fuel — and at every step of conversion across the value chain we are embedding energy from the planet into a product or service, but we are losing energy in this conversion at every step. Aggregate Efficiency is the ratio of the potential to the useful work.” According to Rifkin, we started the Second Industrial Revolution with 3% Aggregate Efficiency, meaning, 97% of every conversion experienced during the value chain — from marshaling resources, shipping them, storing, producing, distributing and recycling them, in every step 3% got into the product and service and 97% was lost. By the 80’s we got to 13% and stayed there until now.
So if we stay in the Second Industrial Revolution general purpose technology platform, productivity is going to continue stale. In his view, we are starting to see the begging of the Third Industrial Revolution. A new convergence between communication, energy and transportation logistics in emerging.
The communications internet (5G) is converging with renewable energies (Solar and Wind) and automated, GPS-guided and soon driverless transportation. This convergence of three technologies he calls the Internet of Things, where connectivity is key and will be managed by Artificially Intelligent ecosystems that form the basis for what economists agree to be a digital economic revolution, or an Intelligence Revolution. This Internet of Things will allow us to bypass all the inefficiencies of the traditional vertically integrated value chains of the Second Industrial Revolution and engage horizontally, without the need for a central controller, at near zero marginal cost, moving our aggregate energy efficiency from 13% to 40% in the next 20 years.
This will free up some goods and services from the capitalist market and make them shareable in the new sharing economy on the collaborative commons”.
In the collaborative commons people share what they have created for the benefit of each other and society. Open source technologies are emerging in horizontal, peer-to-peer networks. Not only in the form of open source software code like Linux, Python and R but also in the Internet of Things.
In the new sharing economy, Millennials prefer access vs ownership. The biggest taxi company in the world, Uber, does not own any cars. The biggest hotel company in the world, AirBnB, does not own any hotel rooms. 3D Printers can also be shared as a common good by members of the commons in a very cost effective way. A start up company called makexyz allows people to share their 3D printers with members of their local communities, like AirBnb for 3D printing. In fact, in Israel, the kibbutzim are collective communities first established in 1909 (Degania) traditionally based in agriculture. Today farming has been supplemented by high-tech and industrial plants. It is a self organized alternative way for the system to distribute labor and capital yielding new forms of value creation and exchange where the means of production are shared by the community.
This new way of cooperation gave rise to a whole new asset class.
A Cambrian explosion of cryptocurrencies species was formed and emerged from the digital currency explosion. This was not only possible due to the blockchain, which ledger system is a distributed network technology by design, where all participants validate each others transactions, but also in the way digital coins are mined, also in a peer-to-peer decentralized and distributed way.
The energy grid is also going peer-to-peer and horizontal. Mini-solar grids are being deployed all over the world giving prosumers (producers/consumers) the ability to produce their own energy and offer the surplus back to the grid. The cost of producing solar panels is also going thru a version of the Moor’s Law known as Swanson’s Law. Swanson’s law is the observation that the price of solar photovoltaic modules tends to drop 20 percent for every doubling of cumulative shipped volume. At present rates, costs go down 75% about every 10 years. Eventually, after fixed costs have been amortized, the marginal cost of producing energy will be near zero. This is a major feat for humanity. That said, time is of the essence as climate change is happening in full force. Reports shows that solar and wind will increase their combined share of global generation capacity to 16% from 3% by 2030. The key driver will be utility-scale solar power plants, as well as the vast adoption of rooftop solar arrays in emerging markets lacking modern grid infrastructure, giving them an opportunity to leapfrog and catch up with developed countries. In places like Latin America and India, the lack of infrastructure will actually make rooftop solar a cheaper option for electricity generation.
The technologies of the future have arrived. Soon, anything you can think of can be build at very low cost. How we are going to utilize these new technologies? What are the most challenging problems we can be tackling now so we can leave the planet better than how we fond it? It’s never too late. Now is the only answer.
(Excerpt from my forthcoming book “An Economic Journey into the Future”)
[1] Durlauf, Steven [1997]-”Trying to understand the behavior of this monster …”-Santa Fe Institute Bulletin, Volume 12, Number 1.








