//Briefing: Economic Complexity – of Robots, T-Shirts and Iron Ore

Briefing: Economic Complexity – of Robots, T-Shirts and Iron Ore

Of Robots, T-Shirts and Iron Ore


Have you ever wondered what goes into the making of a t-shirt, business shirt or blouse – like the one you are most likely wearing right now?

Obviously there are the raw materials – cotton most probably, some plastics perhaps – but what else is important when making a piece of clothing? Most probably the shirt was made with a sewing machine; in a factory; which sat at the end of a road-way or a train-line enabling the raw materials to arrive at the factory gate, and the shirts to leave for sale; the factory probably ran on electricity; the workers were trained to use the sewing machines, in the company culture and values; whilst other workers generated the design and plans for each shirt; whilst still other workers focused on efficient management, incentive schemes, productivity improvements, finance, risk analysis, marketing, sales and surveys.
As you can see, there is a lot going on to make the humble t-shirt.

But what about something more complicated, like a production-line robot? Or something less complicated to find, but just as complicated to make commercially available like iron ore? Robots need similar inputs to t-shirts – labour, raw materials, designs, electricity roads and so on – but they need different raw materials, and more precise scientific knowledge. Iron ore needs mining equipment, engineering knowledge, and of course, an endowment of iron ore to discover and extract.

Now consider the fact that t-shirts, robots and iron-ore are not made in the same place, by the same country. A likely source for your shirt is Pakistan, whilst the manufacturing robotics comes from Germany, yet Australia is the home of global iron ore production. It is worth pausing to consider why should that be? Why should some countries produce only some goods or services to the exclusion of others?

To the Economic Sciences, the observation that different countries specialise in different products is nothing new. For Adam Smith, specialisation, or its partner terminology, the ‘division of labour’ was the key source prosperity for well-governed societies. In his famous treatise on prosperity, the `Wealth of Nations’, he writes,

“It is the great multiplication of the productions of all the different arts, in consequence of the division of labour, which occasions, in a well-governed society, that universal opulence which extends itself to the lowest ranks of the people.”[1]

Later, the Austrian economist, Frederick von Hayek would hit on a similar theme – noting that civilisation is built up not by the training of an individual to hold in their head at one time everything that is known about a production technology or product, but rather, by the combination of distributed information, techniques and inputs. For him, it was count of ‘important operations which we can perform without thinking about them’ [2] coupled to institutions of exchange that made for real advances in prosperity.

Together, Hayek and Smith are pointing to a fundamental, classical insight of the Economic Sciences – nations, and for that matter, all the peoples of the Earth, prosper, when they combine their resources, technologies, products and services in a phenomenally elaborate symphony of exchange.

Perhaps surprisingly, up until very recently, it was extremely difficult for Economists to get a clear handle on the capabilities and complexities of the world-wide division of labour. Trade data is an obvious place to start, but previous methods of analysis have focused on the country to country effects, rather than what the interactions amongst countries can actually tell us about the products and capabilities that enable them.

Recently, a collaboration between physicist, César Hidalgo and Economist, Ricardo Hausmann has seen the development of fascinating techniques to do just that – to study not just the trading relationships amongst the nations of the Earth, but the product relationships that lie behind this trade. Such analysis leads to some tantalizing prospects like mapping capabilities, or measuring the product- or capability- complexity that each country possesses. In this pod-cast, we’ll introduce and explore this remarkable analysis which goes under the heading: Economic Complexity.

Mapping Earthen Production: The Product-Space diagram

Before we continue, we need to outline some key terms in the economic complexity literature. We will use this terminology throughout this article, to help us better understand the nature of each country’s productivity.

  1. First, the diversity of a country refers to the number of different products a single country produces – for instance, if Australia really were a ‘banana republic’ producing only bananas, we’d have very low diversity;
  2. Second, the ubiquity of a product refers to the number of other countries that export a single good – for instance, textiles are exported by dozens of countries and are therefore considered to be ubiquitous, whilst iron-ore is not; and
  3. Finally, a capability refers to an ability or skill required to produce output – German manufacturing robotics relies on Germany possessing robotics and mechatronics capabilities.

Now we know our basic terminology, we can begin to answer the questions we pondered above. The first step in this process is to consider what products require similar capabilities. To do this, we employ a simple trick: we draw a connecting line between any two products that are likely to be produced by the same country. The result can be seen below (Figure 1).

FIGURE 1: 'Product Space' - each node represents a different export product, the colouring indicates the category of product; nodes are linked if the two products are very likely to be exported by the same country -- indicating that the production of the two products requires similar capabilities.
FIGURE 1: 'Product Space' - each node represents a different export product, the colouring indicates the category of product; nodes are linked if the two products are very likely to be exported by the same country -- indicating that the production of the two products requires similar capabilities.

Let’s discuss this further. Consider Pakistan. We know they produce T-Shirts. It makes sense that if they have the knowledge to produce T-shirts, they probably have the knowledge to produce blouses. Or trousers. Upon examining Pakistan’s economy, it becomes clear that they, along with almost every other country that exports T-shirts, also produce blouses, and trousers. T-shirts is therefore represented by one of the blue nodes in the middle of the network: it is connected to many other nodes based on the fact that T-shirts and other textile goods are often commonly produced by the same nation. From this, we can infer that products with many connections require similar capabilities to be produced.

Now consider Australia, or Germany. Australia produces a lot of iron ore, but not everyone has iron ore deposits to export. Germany produce high technology robots, but not everyone has the knowledge or the capabilities to export such products. These goods are not likely to be co-exported with any other goods; hence they are represented by the outlying nodes in the product-space network.

A product’s location in the product-space network can reveal to us how rare a country’s capabilities are. Many countries have the capability to produce textiles, so textiles are located in the centre of the network. We can infer that countries that export textiles have commonly available capabilities. On the other hand, not many countries have the capabilities to produce iron ore or robots; hence these were located on the outskirts of the network. We can infer that countries that export these goods have relatively unique capabilities.

The product-space network has given us a visual representation of which products require more unique capabilities than others…but so far, we have not yet devised a metric which will help us determine which countries produce what goods. In the next paragraph we will develop the concept of economic complexity, and see if this will help us determine the divisions of labour within the world economy.

Measuring Capabilities: Economic Complexity

Economic Complexity: What is it?

We have finally arrived at the punch-line of the article: what is economic complexity? Complexity theory examines how people in an economy interact and combine knowledge, in order to produce output. As Smith and Hayek so eloquently explained to us above, people specialize in order to maximize their productivity. But in order to produce output, people need to put together all their individual pieces of knowledge. For example, to build a house, one needs (amongst other things) an architect to design it, an engineer to ensure it is safe, and a construction worker to actually build it.

To expand on this idea, for an economy to produce output, it relies on interactions between people with complementary capabilities. The more interaction there is, the more knowledge is being shared. And the more knowledge that is being shared, the greater the output an economy can produce. Economic complexity is a measure of the amount of interaction that takes place within an economy.

Differences in economic complexity reflect differences in non-tradable capabilities. For example, poor infrastructure, or weak property rights will reduce the number of interactions that take place between people within the economy. Low labour skills or low technical knowledge will reduce the capabilities a country possesses, thus reducing the number of interactions between people with different capabilities. In essence, economic complexity captures the capabilities of a country AND the interactions of people to combine these capabilities. The higher the number of interactions, the more diverse the range of capabilities must be. And diversity of interactions leads to novelty, attainment of efficiencies of production and so, is a source of economic prosperity.

Economic Complexity: How does it work?

So how do economists currently measure economic complexity? Economic complexity is a combination of the concepts we introduced above: diversity and ubiquity. Let me elaborate, by considering some examples. First, let’s consider Germany. Recall, Germany exports robots. But, upon further examination of Germany’s exports, it becomes apparent they export a diverse variety of goods. Additionally, of the goods they export, very few are co-exported by other countries. Using complexity jargon, German exports are highly diverse and its export products are not ubiquitous. In other words, Germany specializes in products that are hard to produce, which leads us to conclude Germany’s production process is a combination of unique – difficult to replicate – capabilities, which requires a complex web of interactions. This translates to a high ranking on the Economic Complexity Index (ECI).

Let’s compare this to Pakistan. Pakistan specializes in T-shirts, or textiles. In fact, they specialize in little else, meaning their economy is not as diverse as Germany’s. Additionally, much of what Pakistan exports are exported by similar countries. To use complexity jargon, Pakistan’s exports are ubiquitous (easily made by others), yet Pakistani production is not diverse (they only produce a few things). This translates to a low ranking on the ECI.

Finally, let’s consider Australia. Australia specialized in iron ore. As with Pakistan, Australia specializes in little else, leaving our economy relatively un-diverse. But, very few other countries export iron ore, meaning Australia’s exports are not ubiquitous. This presents a conundrum…an economy that isn’t diverse, but very few other economies can produce what Australia does. Does this mean the Australian economy is a complex web of knowledge and capabilities that other countries cannot follow? Unfortunately, no. Let us consider other countries that export iron-ore. If we examine these countries, it is apparent that their other exports are very ubiquitous. In other words, despite the fact they export unique iron-ore, they lack the capabilities to export other unique products. We can only conclude that iron-ore production in not the result of high level capabilities, and a deep network of knowledge sharing, but rather, good fortune. Australia is sitting on volumes of iron ore deposits, while Germany, Pakistan, and many other countries, have none. When taking this into consideration, it is clear Australia’s booming economy does not possess unique capabilities, nor does it require a complex web of interactions and is accordingly given a low ranking on the ECI.

Thus, it is here that we note that the ECI can be a powerful tool for development Economists since it can capture what we intuitively know to be true – that economic development (or rather, economic complexity) is more than just trading in large volumes; true development is about having the conditions which support complicated networks of exchange.

In other words, highly complex economies can weave vast amounts of knowledge together, across large networks. An economy that successfully conveys knowledge through the appropriate networks will prosper and grow. To put this into terms you may be more familiar with, neoclassical jargon would refer to a complex economy as one with a high human capital and good institutions, which leads to economic growth. Conversely, a less complex economy cannot combine knowledge in the same way a complex economy can. The actors in the economy either do not have the knowledge to share, or, if the knowledge is present, the actors face high interaction costs that prevent the linkages of capabilities to arise to produce new products. In short, we expect complex economies to be advanced economies with high income levels, while we expect non-complex economies will suffer from low income, and low economic growth.

Applying the Map: The future of Development?

Now we have our theory, it’s time to test some of our hypotheses. Unsurprisingly, we find that economic complexity is correlated with GDP. But we find the results go deeper than this: economic complexity is a driver of economic growth. Economic Complexity shows us there are two ways economies can grow: they can expand their interactions and find more uses for existing capabilities (the fast way), or they can accumulate new capabilities (the slow way). Applying this logic to the data, we find that countries with a high complexity ranking relative to their income will grow faster than countries with similar incomes. The intuition behind this is that countries with a relatively high complexity have all the tools in place to facilitate economic growth. In other words, the capabilities exist; it is just a case of putting them all together in the right way. Compare this with a country that has a relatively low economic complexity ranking: in order to grow, they need to accumulate new capabilities; a far slower and more difficult task.

By now it should be clear economic complexity has a role to play in economic development, so let’s compare it to other measures used to explain cross-country differences in income. Neo-classical economists have long argued that human capital, physical capital, governance and institutions are all variables which can explain differences in economic growth. However, complexity theory is superior to all these variables when explaining differences in economic growth.

The graphs below demonstrate the strength of Economic Complexity as a predictor of economic growth.

FIGURE 2: The correlation between growth in prosperity (adjusted for initial incomes and natural resource exports) and Economic Complexity; Australia (AUS), Pakistan (PAK) and Germany (DEU) are indicated in the figure.
FIGURE 2: Contributions to variation in Economic Prosperity by various, commonly tested, economic factors showing the powerful correlation that Economic Complexity has with prosperity.

Figure 2 shows that economic complexity explains 73% of variations in cross-country per capita incomes, after controlling for natural resource income. This supports the argument that Economic Complexity plays a powerful role in supporting economic growth.

FIGURE 3: Contributions to variation in Economic Prosperity by various, commonly tested, economic factors showing the powerful correlation that Economic Complexity has with prosperity.
FIGURE 3: Contributions to variation in Economic Prosperity by various, commonly tested, economic factors showing the powerful correlation that Economic Complexity has with prosperity.

Figures 3(a) and 3(b) demonstrate Economic Complexity’s superiority over other growth indicators. Good governance and high institutional quality have long been thought of as important indicators of economic growth, but as we can see in Figure 3(a), Economic Complexity is a far more thorough indicator of economic growth. Economic Complexity accounts for 15.1% of variations in economic growth, while several other governance measures, as used by the World Bank, can account for only 1% of variation in economic growth. Just to clarify: this does not mean that governance and institutional quality is unimportant when explaining economic development. Recall that a high Economic Complexity ranking means a country has the infrastructure in place to promote a high level of interactions, which should reflect a high level of governance and high institutional quality. What this does indicate is that Economic Complexity captures the effects of good governance and institutional quality better than the World Bank measures do.

Another variable that has long been used to explain economic growth is human capital. However, as we can see in Figure 3(b), Economic Complexity has far more explanatory power than any other measure of human capital. Economic Complexity can explain 17.2% of variations in economic growth, while schooling and cognitive ability, two common measures for human capital, account for just 3.6% of the variation. As above, this does not mean that human capital is unimportant when explaining economic growth. A high Economic Complexity ranking means a country has a high number of capabilities, which reflects a high level of human capital. What this shows is the Economic Complexity captures the effects of human capital better than other measures.

From the above data, we can conclude that Economic Complexity may well be the heart of future economic development.

So what does the future hold?

What does economic complexity tell us about the respective positions of Australia, Germany and Pakistan? Australia, complexity tells us, is living well above their capabilities. Australia’s concentration in the mining sector has left us with skills shortages in other areas. If (when) the mining boom collapses, Australia will find itself with high levels of structural unemployment, as the economy will require re-training. Policy makers should be looking to reeducate the Australian workforce, in order to ensure Australia has a diverse range of skills in the future.

Pakistani policy makers face several issues. Not only does their workforce require education and diversification, they also need to ensure that there are sufficient avenues to transmit and utilize this knowledge. There needs to be public expenditure on infrastructure, in order to ensure networks can be created in order to combine new capabilities.

Finally, Germany is in a position of strength. They have the systems in place to facilitate the transfer of knowledge between agents, and they have an educated and diverse labour force. The challenge for Germany now lies in finding new avenues to economic growth. For this to happen, Germany must focus on developing even more capabilities, as they have most likely they have discovered all the uses of their current capabilities.

Credo: Of Cars, Cash, and Capabilities

In Australia, we may already be seeing the roots of complexity theory take hold in some policy decisions. For example, the rising Aussie dollar has put extreme pressure on the manufacturing industry. Producing manufactured goods in Australia has become costly and car makers, such as Holden, have been looking to move their production overseas. The government, fearing the massive loss of jobs that may occur if Holden was to cease production in Australia, has recently given a $275 million bailout package. [3] Now, first year undergraduate economics would tell us that the government has made a bad decision: if Australia are inefficient at producing cars, then Australia should get out of the manufacturing market, and move into mining, which is where our comparative advantage lies.

However, hopefully Economic Complexity has shed a new light on this argument. Complexity theory has shown us the importance of a diverse workforce, and has outlined the potential costs in losing an entire industry, and with it, an entire capability, especially considering the manufacturing industry produces relatively low ubiquity products. If Australia was to lose the manufacturing industry, not only will this harm the diversity of our exports, we will also be losing some unique capabilities, which may be crucial to support production in several other industries (an argument made often by the Labor movement, though for different reasons). Economic Complexity has shown us that the production process is a complex web of interactions and networks between several complementary capabilities. The entire process may suffer if some of the key nodes of the networks are removed. The question, ultimately, is whether saving Australia’s manufacturing capabilities is worth the $275 million? The next frontier for Economic Complexity may be to quantify the importance of every industry, but until then, only time will tell.


  1. Smith, A. (1993) Book 1, Chapter 1 in An Inquiry into the Nature and Causes of the Wealth of Nations. Oxford University Press. New York. p18
  2. Hayek, F. (2010). Studies on the Abuse and Decline of Reason: Text and Documents. The University of Chicago Press. Chicago. p149
  3. ‘Holden given millions to stay in Australia’, The Age, 22 March 2012, available here.


  1. Hidalgo, C. A., Klinger, B., Barabasi, A. L., & Hausmann, R. (2007). The Product Space Conditions the Development of Nations. Science, 317(5837), 482–487.
  2. Hidalgo, C., & Hausmann, R. (2009). The building blocks of economic complexity. Proceedings of the National Academy of Sciences, 106(26), 10570.
  3. The Atlas of Economic Complexity Observatory at MIT
    Or download the podcast here: download (mp3)

Dr SIMON ANGUS is a computational and complexity scientist and member of the Department of Economics, Monash University. With a background across the physical and social sciences, he has diverse interests including complex systems science, data-science, networks, systems biology, evolutionary game theory and the study of technology.