Tag Archives: complexity

Graphic from

As a follow-up to Eric Beinhocker’s The Origin of Wealth, I recently downloaded and read The Atlas of Economic Complexity by Ricardo Hausmann, Cesar Hidalgo et al. It was a good chaser after Beinhocker’s massive introduction to complexity economics. Hausmann and Hidalgo are influenced by the idea that the economy is a complex adaptive system. Therefore they reject the idea that you can easily sum up an economy in a single number like GDP. Instead, they try to analyze all the different products produced in the economy in an attempt to get a grasp on how complex it is rather than just how big it is in dollar terms. Their results are interesting.

Hausmann and Hidalgo assume that the economy is based on productive knowledge–bits of knowledge necessary to make the products that we consume. These units of productive knowledge they call capabilities. They also assume that many products have overlaps in the capabilities needed to make them. So if you make shirts, there is a high probability you can make blouses too. These relationships should show up in a visualization of the product space. Countries that makes shirts will show a tendency to produce products with overlapping capabilities: blouses, pants, etc.

Hausmann and Hidalgo calculate a country’s economic complexity by measuring the diversity of products a country is capable of producing, and by also calculating the ubiquity of the products they do produce. So if a country produces only a few products and those products are ubiquitous throughout the world it is pretty certain the country has a fairly non-complex economy. The amount of capabilities present in that economy must be pretty few. For example, here’s an infographic of the economy of Tajikistan:

Taj exportsAluminum, raw cotton and dried fruit dominate the Tajik economy. And each of these items are ubiquitous enough in the world that it does not take a high degree of scarce knowledge to produce them. On the other hand, here’s Thailand’s exports:

tree_map_export_tha_all_show_2010Notice the increase in the number of products, but also of the kinds of products produced. It takes many more capabilities to produce electronics than it does to produce raw cotton. Here the difference between an emerging economy like Thailand and an underdeveloped economy like Tajikistan is pretty stark.

The interesting thing about the measure of economic complexity is that Hausmann and Hidalgo have found it to be a strong indicator of future economic growth. The economies which are highly complex but with a lower-than-expected current GDP can be expected to grow quickly, while countries that have a high GDP relative to their complexity can be expected to grow slowly, if at all. This gives a new dimension to the “resource curse” hypothesis in that it displays growth based on natural resource exploitation is unsustainable, unless it is invested in expanding other productive capacities.

According to the Atlas of Economic Complexity, the economies whose level of complexity  most predict growth in GDP per capita are China, India and Thailand. Next in the rankings come Belarus, Muldova and Zimbabwe. All of these countries have economies which currently lag behind their potential.

Using GDP instead of GDP per capita, the top slots all go to Sub-Saharan Africa: Uganda, Kenya and Tanzania take the top slots, with Zimbabwe, Madagascar and Senegal following. Though their high levels of population growth keep income per capita down, these will all likely be fast-growing economies over the next decade.

The Atlas of Economic Complexity brings to life the incredible diversity within the world economy. It offers a new metric of development: the Economic Complexity Index, or ECI, which may prove to be a more important indicator than more simplistic metrics like GDP per capita. Time will tell how Hausmann and Ricardo’s predictions turn out. But their approach seems bound to be imitated as development theorists absorb and make use of the insights of complexity science.

For a brief introduction to economic complexity from Cesar Hidalgo himself, check out his talk at TEDx Boston:

And for more cool data visualizations visit the Observatory of Economic Complexity.


Every once in a while I read a book that totally revolutionizes my conceptual categories. Eric Beinhocker’s The Origin of Wealth: Evolution, Complexity, and the Radical Remaking of Economics did that form me in the field of economics.

Some background: economists have known for some time that there are some pretty big caveats to the models we’ve been using to understand the economy. These models are hugely contingent on some key assumptions including:

-the rational decision-making of (perfectly) informed economic agents.

-the tendency of markets to reach equilibrium, unless shocked from some exogenous source

Unfortunately, in the real world, these assumptions are largely wrong. And recent events, including the Great Recession and euro-zone crisis–both of which were not foreseen by most mainstream economists–have significantly increased the skepticism with which many view the models that come from these assumptions.

But what is the next step for economics? Behavioral economics has long been punching holes in the vision of the rational, perfectly informed agent. But the mainstream view is still of an equilibrium economy, with caveats for market failures. There has been little in the way of a new synthesis.

But Eric Beinhocker thinks it’s coming. His book advocates a bundle of related insights he  calls collectively “Complexity Economics.” I would venture to bet these insights will be the source of economics’ new mainstream.

Complexity Economics, as Beinhocker frames it, arises out of the mathematical revolution that brought us chaos theory. This is the mathematics of unpredictable, turbulent, emergent behavior, sensitively dependent on initial conditions and subject to amplifying feedback loops that defy strictly Newtonian attempts at analysis. This is the world of turbulence, fractals, emergent intelligence, population dynamics and a whole host of other phenomena which math had no tools to understand until the 1980s. If you want a readable introduction to this stuff, I recommend James Gleick’s Chaos: Making a New Science.

These mathematical innovations are finally coming home to roost in economics, the study of the most massively complex emergent system humankind has ever created. And Beinhocker’s book is a manifesto of sorts for this new approach.

So what is Complexity Economics? Beinhocker distinguishes it from what he calls “Traditional Economics,” (his term for pre-complexity equilibrium economics) using the following categories:

Dynamics: Traditional Economics tends to depict “closed, static, linear systems in equilibrium.” Complexity economics favors “open, dynamic, nonlinear systems, far from equilibrium.”

Agents: Traditional Economic agents are “modeled collectively” and tend to “make no errors and have no biases; have no need for learning or adaptation.” On the other hand agents in Complexity Economics are “modeled individually” (this matters because their behavior gives rise to macroeconomic phenomena in an emergent manner, rather than according to immutable laws of supply and demand as in Traditional Economic theories) and “are subject to errors and biases; learn and adapt over time.”

Networks: Traditional Economic models “assume agents only interact indirectly through market mechanisms (e.g., auctions),” while Complexity Economics “explicitly models interactions between individual agents; networks of relationships change over time.”

Emergence: In Traditional Economics, “Micro and macroeconomics remain separate disciplines,” while in Complexity Economics “macro patters are the emergent result of micro-level behaviors and interactions.”

Evolution: Traditional Economics offers “no mechanism for endogenously creating novelty, or growth in order and complexity.” Complexity Economics proposes an “eveolutionary process of differentiation, selection, and amplification” as an explanation for the process of economic development.

Taken together, these distinctions create a whole new way of looking at economics. Economists tending toward this school of thought tend to create models of individual agents, and then watch what structure emerges–like a whirl in a turbulent stream–from their interaction. This is in contrast with typical models that assume everyone is totally self-interested and perfectly rational, then create laws that bring the model to an equilibrium where utility is maximized in the system–more like a marble coming to rest in a bowl.

For those on the edge of the field, this change has a few important implications:

1. The market is neither rational nor predictable. This has been well-demonstrated by the past 100 years of economic history, but until now we haven’t had the math to describe a non-equilibrium, non-linear, complex adaptive market. Now we do.

2. The market is adaptive, as are the agents in it. Previous thinkers have compared the market to the survival-of-the-fittest world of nature. But now economists are actually modeling firms and business plans as if each is subject to the same selection pressure as a bacteria in a competitive ecosystem. In other words, each firm is competing for energy. Literally. Every firm demands energy in some form, and uses it to create useful order. How useful the orderly stuff is determines whether the firm gets more energy to keep creating useful, orderly stuff. This is not a metaphor. The complexity models are literally placing economics within the constraints of the second law of thermodynamics. Either a firm adapts, or it is collapses into high entropy.

3. We might not be able to predict the economy, but we may be able to give it boundaries. Though the economy is a shape-shifting emergent phenomena involving billions of agents and trillions of goods, like an eco-system, it has its boundaries. There is a “fitness function” to which all firms conform: the demand of consumers. And if we, as flawed but self-conscious actors in this complex system, demand (either through uncoordinated action, or political will), an economy whose effects on the environment and the poor are less deleterious, we may get it. But it will be difficult, and our actions will be plagued by unintended consequences.

The Origin of Wealth gave me a lot to think about. And its titular proposition: that wealth is the result of low-entropy order emerging from social conditions favorable to the fostering of complex, dynamic systems, is an interesting one.

It dovetails with Why Nations Fail, which I wrote about some time ago. But while Acemoglu and Robinson could be frustratingly mono-causal in their discussion of political institutions, Beinhocker is frustratingly vague on the actual mechanisms for creating the conditions required for complex, dynamic, wealth-creating systems to get working for humanity. But, as Beinhocker readily admits, this is a new turn for economics.

Perhaps more insights will be forthcoming as the field progresses. A number of development thinkers I follow, including Oxfam’s Duncan Green and Owen Barder of the Center for Global Development, have begun writing about the implications of complexity theory for development, and Ben Ramalingam has a fascinating blog compiling research from the intersection of aid and complexity.  I’ll likely write more as I learn about it.