Ricardo Hausmann on Economic Complexity

The member meeting at the Media Lab features speakers from within the lab, like César Hidalgo and Joi Ito, and outside speakers – in that latter case, the invited speakers reflect César’s wonderfully idiosyncratic take on networks. One of his major collaborators is Ricardo Hausmann, director of Harvard’s Center for International Development and former Minister of Planning for Venezuela.

Hausmann argues that to succeed economically, humans have learned how to specialize. Someone who’s marvelous in one area is likely mediocre at others – consider Michael Jordan’s ill-fated attempts to play professional baseball. Some tasks require a full human’s worth of knowledge – a person-byte – to carry them out successfully. Others require much more knowledge – building a complex product like a computer might require a kilo-person byte or more – the highly specialized knowledge and skills of a thousand different people. “Modern man is useless as an individual. Making a computer is a team sport.”

By understanding how much knowledge and coordination different economies are capable of, we might understand their economic growth potential. In the US, the average employee works with 100 coworkers. In India, the average employee works with 4 coworkers. Hausmann explains that’s not coincidental – the difference in wealth and income between the nations is closely related to the ability of firms to take on complex tasks. This also helps explain recent disappointment with the limited impacts of microlending – those loans go to small firms that are limited in terms of personbytes. They’ve only got so much knowledge they can apply to producing complex and high value products.

We might characterize economies in terms of those where lots of people do very simple work – he illustrates this with a marvelous Edward Burtynsky photo of assembly line workers processing chicken in China – and those where indiviuals do complex things in consort, like the players within a symphony orchestra. Hausmann shows us a “map” of the world, a complex graph that represents nations and what products they produce. Most nations produce a few things, and a few produce many different things. Some products are made everywhere, while others are made in very few places.

There’s an underlying pattern to this. The nations that make only a few things all tend to make, more or less, the same things. Basically, we can divide the world into two sets of countries – those that have sufficient personbytes of knowledge to produce a wide range of goods, and those that can produce only a few simple things. The places that make everything make things that few others make. Hausmann explains that products require a specific set of personbytes to produce. When you gain additional personbytes of skill, it’s like getting new letters in Scrabble – you can produce a new set of words, but only within the constraints of the letters (skills, knowledge) you already have.

“Poor countries make few things, and things that everyone makes. Rich countries make unique things. And this is true for municipalities as well as for countries.” He shows a graph of manufacturing in Chile that looks curiously like his graph of the world – on the top is Santiago, where people manufacture all sorts of things… on the bottom “is where there’s nothing but penguins” and capacity for manufacturing is very low.

Global economics, Hausmann explains, is a little like the BCS scoring in college football. It’s not just about who you beat, it’s about who they beat as well. What do you make, and what does everyone else make? What do you make that no one else makes? What new products could you manufacture based on what you already make?

Why pay attention to this idea, the “economic complexity index”? It’s a very good tool for explaining the classic question of “Why are some countries rich and others poor?” Specifically, it explains 73% of the variances of incomes across nations. And where the predictions economic complexity theory offers differ from reality, it’s possible that reality is wrong. The index suggests that India should be richer and Greece should be poorer, which suggests that error in the index is predictive of future growth. If you want to bet on economies that are undervalued, Hausmann suggests you invest in China, India, Thailand, Belarus, Moldova and Zimbabwe. (On the last, he suggests that Zimbabwe’s main economic problem is a single persistent individual, but that there are many personbytes of knowledge ready to produce goods once the political situation changes.)

Is economic complexity actually measuring another phenomenon, like education? Probably not. We can look at investment in education and economic growth, and education appears to correlate more weakly than economic complexity. He suggests we look at Ghana, which has invested heavily in education since 1975, and Thailand, which hasn’t invested as heavily. Ghana hasn’t moved far from a largely agricultural economy, while Thaliand has moved from producing jute and sugar to becoming a major manufacturing center. They’ve accumulated many personbytes even if they didn’t invest heavily in education.

This raises a tricky question – how do you become a watchmaker in a country without watchmakers? The answer is that you move from what you currently produce to products that require only a fractional increase in personbytes, from one product space to a closely related one. The question for economic success may be how close you are to good products from what you already know how to make.

I find Professor Hausmann’s theory fascinating, in part because I’ve had the chance to play with the gorgeous visualizations César has built of economic progress in different parts of the world based on economic complexity. What I still don’t understand is how Thailand kicked Ghana’s butt economically. How do you get from jute to microcircuitry? And why couldn’t Ghana get from aluminum production to more complex manufacturing. Looking forward to reading his papers and understanding a bit more, as the core concept of complexity is a very compelling one.