Tim Hartford describes some very interesting work being done by physicists and economists to map out the relationship between products at global and national scale (hat tip: Marginal Revolution). The idea seems to be that poor countries have different product clusters (say, tropical agriculture and mining) from rich nations and it's not easy moving from one to another.
"The physicists' map shows each economy in this network of products, by highlighting the products each country exported. Over time, economies move across the product map as their export mix changes. Rich countries have larger, more diversified economies, and so produce lots of products—especially products close to the densely connected heart of the network. East Asian economies look very different, with a big cluster around textiles and another around electronics manufacturing, and—contrary to the hype—not much activity in the products produced by rich countries. African countries tend to produce a
few products with no great similarity to any others."
That could be a big problem. The network maps show that economies tend to develop through closely related products. A country such as Colombia makes products that are well connected on the network, and so there are plenty of opportunities for private firms to move in to, provided other parts of the business climate allow it. But many of South Africa's current exports—diamonds, for example—are not very similar to anything.
If the country is to develop new products, it will mean making a big leap. The data show that such leaps are unusual.
This is a pretty compelling argument for industrial policy (paging Dani Rodrik!). The question, of course, is what type of industrial policy is best and how it can be implemented. Needless to say, this debate has been going on for nearly 50 years and, curiously, but it's striking how little we know.