Genetic diversity is an interesting phenomenon. Too little of it and the population can be too homogeneous, too much of it and there will be a lot of mistrust and conflict. This exactly sounds like something that could potentially influence economic development, doesn’t it?
Research on genetic diversity’s effect on development is quite new. It started with Ashraf and Galor’s (2013) paper, which established the pattern described above: that there is a hump-shaped relationship between genetic diversity and development. Some remain skeptical, however. So let us look at a new take at this question.
A new working paper by Ager and Bruckner (2013) looks at how different genetic diversity in U.S. counties affected subsequent long-term economic development. The authors take advantage of the fact that around 1870, lots of mostly European immigrants came to the United States. These newcomers affected the genetic diversity of the counties they settled in. Ashraf and Galor’s (2013) theory would then predict that county-wide growth rates were affected by this. Specifically, it would predict that counties that received little immigration, and consequently remained genetically homogeneous developed more slowly than counties that received an optimal amount of immigration, and consequently an optimal boost in genetic diversity. Furthermore, counties that received a large amount of immigration, and consequently became genetically fractionalized should also have developed slowly.
Ager and Bruckner (2013) take advantage of U.S. Census data from 1870 documenting the country of origin of each resident. Ashraf and Galor (2013) collected data on the genetic distance from East Africa of each country. Thus the authors can use this data, weighted by the number of immigrants from each country, and calculate genetic diversity for each U.S. county in their sample. They regress this measurement of county-level genetic diversity, and various control variables such as initial GDP or immigration levels on county GDP growth in the period 1870-1920 (which is when the mass immigration ended).
The results seem crystal clear. Genetic diversity has a positive, highly significant (p < 0.01) effect on a county’s GDP growth. This result is verified by a large number of robustness checks.
Note though that the results indicate a positive, linear relationship between GDP and genetic diversity, as opposed to the hump-shaped relationship proposed by Ashraf and Galor (2013). The reason for this is simply that the overwhelming majority of immigrants were of European origin. Europe has moderate genetic diversity. Thus these immigrants could not have possibly brought too large amounts of genetic diversity to the U.S. They just brought a moderate amount. Therefore, a hump-shaped relationship cannot be observed, but the results are still not contradictory to those of Ashraf and Galor (2013).
Quantitatively, a one standard deviation increase in genetic diversity increased GDP growth by around 21%, or 0.4% per year in the period 1870-1920. The authors also look at more long-term effects, and find that 1870 genetic diversity still has a significant positive effect on 2010 GDPs.
One problematic thing in this paper is the authors calculating county-level genetic diversity by simply averaging each group’s genetic diversity weighted by the group’s size in the county. Although I’m not a biologist/geneticist, but as far as I understand genetic diversity this is a wrong reasoning. Consider two groups, A and B, both with a genetic diversity of 0.65. The authors assume that if you put these two groups together in a county in any mix, that county’s genetic diversity will be 0.65. But what if group A’s diversity mostly comes from varying external characteristics (such as differing height, weight, body types, looks), whereas group B has mostly homogeneous external characteristics. Their diversity rather comes from varying internal characteristics such as personality. Putting these two groups together would then yield a group that is even more heterogeneous than either group A or B alone.
This is exactly why when Ashraf and Galor (2013) construct their measure of post-colonial genetic diversity, they don’t just calculate weighted averages, they also take the differences between all the subgroups in a country into account.
Secondly, in Ashraf and Galor (2013) genetic diversity is a proxy for cultural and other diversity within a group. And its effects on development are measured on the scale of centuries and millenia. The authors here mainly use a time span of 50 years. Can any kind of diversity really act so fast on development? I have my doubts. Diverse vs. homogeneous communities may on average have different development paths, but it takes time (read: centuries) for this difference to show.
So then what accounts for the results? Well, it’s hard to tell but probably the authors neglected to control for some important determinants of growth. Potentially, immigrants went to already rich counties, or to those that were full of potential. These counties were probably in close proximity to cities, probably had higher population and population growth in general, potentially had better geography and natural resources. They probably also had a higher population density, better infrastructure and better rule of law.
Also, if genetic diversity correlates with development in Europe, then it could be that immigrants from more successful European countries (i.e. immigrants with higher diversity) settled down in counties that they subsequently made more successful. Immigrants from poorer European countries did not make that large of a positive impact on the counties they settled down in. This could be due to differences in skills, language, culture, or that these people were used to extractive institutions in their home countries, which altogether altered their behavior. Another reason could be that it was mostly adventurous people with an entrepreneurial spirit who migrated from rich countries, whereas from poorer countries a broader group emigrated simply because of larger poverty. Controlling for home country development levels could help a lot here.
In sum, because of these problems, I don’t find the paper convincing. Perhaps with better controls and a better metric for genetic diversity, it could be made into a reasonable paper. However, it only shows that moderate levels of diversity have positive effects on development in a democratic setting, a finding that is not really new. It doesn’t confirm the hump-shaped relationship of Ashraf and Galor (2013) on a smaller scale.