The new Oded Galor and Quamrul Ashraf paper

Here is from an editorial summary published in Science (gated):

…Ashraf and and Galor present the hypothesis that genetic diversity has exerted a long-lasting effect on economic development—which is quantified as population density in the precolonial era and as per-capita income for contemporary nations—beyond the influences of geography, institutions, and culture. They posit that intermediate levels of heterozygosity allow for a productive balance between the social costs of high diversity and the creative benefits of higher variance in cognitive skills. They show that the optimal level of diversity was approximately 0.68 in 1500 CE, and that this increased to 0.72 (which is pretty much where the United States sits) in the year 2000, with the most homogeneous country, Bolivia, placed at 0.63 and the most diverse country, Ethiopia, at 0.77. Just how large an effect are we talking about? They estimate that genetic diversity accounts of 16% of the cross-country dispersion in per-capita income; put in another way, shifting the diversity of the United States higher or lower by one percentage point would decrease per-capita income by 1.9%.

One version of the paper is here, and it will be coming out in the American Economic Review.  Being on the road, I have yet to read this work.


"genetic diversity has exerted a long-lasting effect": it's rare that a correlative study can imply causation, never mind the direction of a causation.

Three words: correlation is not causation -- this is just another bogus "study"

+1. Bad enough that we have massive overgeneralization in evolutionary psychology. Now it's spread to economics?

Have you read the paper, or even a small part of it before making these superficial comments? I doubt it. The empirical analysis is very careful. For instance, the authors write: “The present study exploits the explanatory power of migratory distance from East Africa for genetic diversity within ethnic groups in order to overcome the data limitations and potential
endogeneity issues encountered by the initial analysis discussed above. In particular, the strong ability of prehistoric migratory distance from East Africa in explaining observed genetic diversity permits the analysis to generate predicted values of genetic diversity using migratory distance for countries for which diversity data are currently unavailable. This enables a subsequent analysis to estimate the effects of genetic diversity, as predicted by migratory distance from East Africa, in a much larger sample of countries. Moreover, given the obvious exogeneity of migratory distance from
East Africa with respect to development outcomes in the Common Era, the use of migratory distance to project genetic diversity alleviates concerns regarding the potential endogeneity between observed genetic diversity and economic development.”


I have two observations questions after a very brief look at the paper; the story sounds plausible, and they do relate humans to other species, however I have my doubts about these sort of "biological" explanations:

1. If we compare a number of countries where ethnic cleansing took place, we should be able to explain part of the economic growth since then in terms of a change in the genetic diversity. It will be difficult to keep things c.p, but that should be possible correct?

2. What's the difference between this study and the study that investigates the relationship between male penile length and economic growth?

I know, it sounds ridiculous, but the author gives an economic explanation of why it could matter; why is that study more 'serious' than the male organ length hypothesis?

"This study explores the link between economic growth and penile length between 1960 and 1985. It estimates an augmented Solow model utilizing the Mankiw-Romer-Weil 121 country dataset. The size of male organ is found to have an inverse U-shaped relationship with the level of GDP in 1985. Economic development between 1960 and 1985 is negatively associated with the size of male organ. With considerable reservations it is also found to be a more important determinant of GDP growth than country's political regime type. Two interpretations for the patterns between male organ and economic growth are discussed briefly: the link between penile length, testosterone and risk-taking, and selfesteem production. Despite the robust statistical links, until more rigorous treatments on the subject the proposed ‘male organ hypothesis’ should be taken with reservations."

Obviously, the creation of national borders is endogenous to ways in which ethnic violence is restrained by institutions, geography, accidents of history, etc.

Ack, they have a panel with 21 data points and many variables and .70+ R^2, yet they claim *** level of confidence on quite a few variables. Did no one read Student?

The chance for spurious correlations here are really, really high.

The XKCD jelly bean study comes to mind.

How is Bolivia "the most homogeneous country"? It has a prominent permanent racial divide due to altitude differences, with white women having very hard time giving birth above about 10,000 feet. The city of La Paz is notorious for being all Indian in the thin air suburbs, while the wealthiest and whitest part of town is at the far bottom of the canyon.

If you define observed diversity as "The average expected heterozygosity across ethnic groups from the HGDP-CEPH Human Genome Diversity Cell Line Panel that are located within a given country." Obviously!

Genetic diversity among amerindians is the lowest in the world. It doesn't seem unlikely that Bolivia, having one of the largest representation of amerindians in its population (among countries in Latin America), has the lowest level of diversity.

Instead of looking at biological diversity, perhaps you should look at human capital diversity and the costs of homogeneous monoculture with the presence of minorities which are discriminated against, receive less human capital investment than the dominant group.. In other words, if you were to look at the 1860s south, for example, or even the 1950s, you would be mapping the genetics variable when in fact it was a cultural variable associated with a genetics variable that led to outcomes; similarly, and along the same path, with more varied minorities, and the inability of a dominant group to restrict access to human capital investment, you might have a different outcome. Again, you might be mapping genetics as the variable, but completely overlooking the more dominant social aspects of human interaction.

Human interaction and interaction effects are more related to culture than to genetics.

I bet they mispecified the variable.

To see some of this, go back and take the lead paragraph the post, and substitute for genetics or genetic diversity the words "cultural diversity" or "culture". Ask how the increase in the growth in multiple minority cultures increases variability. Similarly, take the periods where there were changes in diversity and look at how societal resources were realocated differently to different elements of society as the composition of society changed.

And how is Ethiopia "the most diverse country"? It's like a reverse image of Bolivia, another high altitude country. Ethiopia is all black in hot lowlands and mixed white-black in the highlands, as a glance at a picture of the late Emperor Haile Selaise would suggest.

If you are looking for the most mixed country, perhaps Surinam or Guyana.

I'm assuming it's diverse because "black" isn't an ethnic group

"I’m assuming it’s diverse because “black” isn’t an ethnic group"

Yes, but neither is brown or white.

Africans have much higher generic diversity than caucasians and other groups.

Africans aren't a group they are a collection of groups. Do Africans have a higher diversity than North Americans, South Americans, Asians? Is that even a good question? How do you measure genetic diversity? Is the way you currently measure it the best way? Why would Africans even be the right terminology when talking about genetic diversity? Etc.

Genetically it is well established that people in Africa are the most genetically diverse humans. You can measure that empirically by looking at the amount of SNPs (single nucleotide polymorphisms) present in a population. If population A has on average 1200 SNPs and population B has 800, population A is more genetically diverse because for any given person there's more sites on the genome that may have variations. The observation that Africans are more genetically diverse is a cornerstone of the "out of Africa" theory - it makes sense that if humans evolved in the Horn of Africa you would expect to see the widest variety of humans there, with humans becoming more and more homogenous the further away you are from the site of species origin (which explains why Native Americans are the most genetically homogenous group). These measures of "genetic diversity" have less to do with what modern ethnic groups are in a specific area than they do with patterns of human migrations hundreds of thousands of years ago.

"They posit that intermediate levels of heterozygosity allow for a productive balance between the social costs of high diversity and the creative benefits of higher variance in cognitive skills."

Higher variance in cognitive skills would have creative benefits because ... uh ... because engineering problems come with different right answers depending upon what your genes are? I don't really think that's how it works.

This reminds me of Yglesias complaining that Apple is obviously leaving money on the table because all twelve of its highest ranking executives are white men.

High variance can be advantageous for two reasons. One, higher variance means more 'super achievers'. The weight above 2sigma will be much larger!

Also: the economy does not consist solely of engineers. By having a high variance population, one can populate all the different careers. In times of economic change, high variance would ensure there is always someone to fill new positions. If everyone's an engineer, and all the openings are for health care, you're going to be having a lot more economic dislocation.

I'm sure there are other reasons, but I'm sleepy.

"higher variance means more ‘super achievers’". Actually, no it does not. If you have group A with higher average, and group B with lower average, than A+B mixture would have smaller % of high achievers than group A alone, despite having a higher variance. Similarly for you second point "By having a high variance population, one can populate all the different careers." - people with high abilities can populate all levels of jobs; people with low abilities cannot.

Ya ya, I understand that high variance can mean bimodality vs just a larger spread in a gaussian. I was assuming that this was an economics paper so by CLT, it's a gaussian =P More importantly, that is how "higher variance" can possibly be advantageous, not that that is exactly what is happening. I'll try to be more pedantic about my assumptions when commenting from now on. Maybe we should assume we're talking about the entropy instead of the variance?

On my second point, high ability can exist in a multi-dimensional space, you're assuming it's basically 1d, yeah? And even if the magnitude of that vector is lower in certain individuals, higher variance (entropy etc) would imply - to me, and this is what I was trying to get at - that we're maybe filling out the 'high dimensional space' of abilities better (arts, engineering, plumbing, etc as concrete though obviously slightly misleading examples), and then we get all the benefits of comparative advantage....

That's what I was trying to get at, if it makes any sense

Read up on IQ - the non-politically correct term for cognitive skillss - a bit. Even truck drivers do better with higher IQ. Here is a good introduction
There is no advantage in having lower IQ for economic development.

Steve, what's the best way to really learn something? To teach it to someone else, right? Well why is that? Because hearing yourself talk makes you smarter? Of course not, it's because your student doesn't start at the same point or process the information in the same way as you. By getting on your students' level and helping them understand, you, the teacher, actually end up learning something too. Less intellectual diversity, less teachable/creative moments. Seems plausible to me.

From the abstract:

"While the intermediate level of genetic diversity prevalent among the Asian and European populations has been conducive for development, the high degree of diversity among African populations and the low degree of diversity among Native American populations have been a detrimental force in the development of these regions."

No, this is a standard misinterpretation of population geneticists' phrase "genetic diversity." People need to finally grasp that population geneticists actively try to avoid looking at genes that do much of anything, because those important genes get selected. Population geneticists look for the most unimportant genes they can find because mutations accumulate in them, allowing the geneticists to track the pedigrees of the population.

Thus, New World Indian populations have a low degree of diversity in their junk genes because they came through a number of bottlenecks to get here, while sub-Saharan Africans have a high degree of junk gene diversity because they didn't go anywhere, so they didn't go through as many bottlenecks. But these genes aren't very important for phenotypes. If they were, population geneticists wouldn't use them.

Just the kind of paper AER would lap up. That journal loves sensational headlines.

Yes, it's quite a fiasco. It's embarrassing for both AER and Science to give so much publicity to a paper based on a simple but fundamental misunderstanding of what population geneticists are talking about when they say "Sub-Saharan Africa has the most genetic diversity."

Here's a bit of advice that would save economists a lot of grief when trying to think about population genetics: pay attention to your lying eyes. Ask yourself: Do Sub-Saharan Africans _look_ like they have huge amounts of genetic diversity?

The answer is: not really. Nilotics look kind of different, and there are small populations of Khoi-San and Pygmies who look quite different, but a huge swath of Africa is populated by people who don't look that different. And that's because they aren't that different genetically. They're mostly descended from the Bantu expansion that came out of West Africa a few thousand years ago and swept over much of the continent.

Granted, there are few reference books published anymore with pictures of what people look like in different places because Knowledge Is Evil, but you can just go to Google Images and look at tourists' pictures on Flickr of what the locals look like.

'Ask yourself: Do Sub-Saharan Africans _look_ like they have huge amounts of genetic diversity? '
Ah yes, the just look at the skin color approach to racialism (to use an apparently favored term among such people, because the older terms apparently are just too accurately burdened with historical reality).

Why point to DNA results when your lying eyes can't even see them?

Because economists don't understand population genetics, so they take cliches passed on by journalists who don't understand population genetics, and they wind up writing papers based on a misunderstanding of what population geneticists mean by "genetic diversity." If they'd consulted their lying eyes, they might have said to themselves: You know, this doesn't match up. Maybe we should read up more on population genetics to make sure we're interpreting this stuff correctly.

"there are few reference books published anymore with pictures of what people look like in different places because Knowledge Is Evil"

There are fewer books about phrenology, but that's because there is a smaller racist market, and more people who dislike racism.

"No, this is a standard misinterpretation of population geneticists’ phrase “genetic diversity.”... Do Sub-Saharan Africans _look_ like they have huge amounts of genetic diversity?"

Steve, the way natural selection can and has interacted with extant genetic variation is an open question, but the greater genetic diversity of Sub-Saharan Africans is not a myth, and it is not limited to some hypothetically useless sub-set of the genome. In fact, black Africans do exhibit greater physical diversity:

"Using a large data set of skull measurements and an analytical framework equivalent to that used for genetic data, we show that the loss in genetic diversity has been mirrored by a loss in phenotypic variability. We find evidence for an African origin, placed somewhere in the central/southern part of the continent, which harbours the highest intra-population diversity in phenotypic measurements."

No no no, you see, all those black people just look alike to me, so, obviously, there's not much genetic diversity.

Steve Sailer makes very insightful comments for someone who is a columnist, film critic and a journalist. The level of reporting would be highly elevated if we have more Steve Sailers. Unfortunately, he is attempting to pass himself as a biologist or geneticist in this comment thread.

What is actually funny is that he is committing the same sin that he is accusing economists of: "Because economists don’t understand population genetics, so they take cliches passed on by journalists who don’t understand population genetics..." The truth of the matter is that Sailer's understanding of population genetics comes almost entirely from what he reads on the GNXP blog, West Hunter blog, a few other specialized blogs, and a discontinued mailing list that he helped set-up a few years ago. I doubt he has personally a read an actual peer-reviewed genetics paper on his own.

It's mind-boggling to me that I can see a comment of yours, post a reply, and in that time you'll have managed to create a whole new post (every time!) that gets sillier and sillier. "aren't very important for phenotypes"? "junk genes"? wtf?

You're not really up to speed, are you?

'You’re not really up to speed, are you?'
What Sailer means to say is you really don't know what he thinks, do you? Just check under 'racialist' - he is one of the new breed, though you might recognize all the old traits associated with previous terms for his style of thinking.

Whatever you think of Steve, his arguments here stand on their own merits.

This: 'population geneticists actively try to avoid looking at genes that do much of anything, because those important genes get selected. Population geneticists look for the most unimportant genes they can find because mutations accumulate in them, allowing the geneticists to track the pedigrees of the population.'

is, I think, a true statement. Are there any population geneticists out there who care to confirm or refute this statement?

Each post like this should have a disclaimer stating "Remember: two out of three published papers' results can't be reproduced. And the odds are even worse for interesting results."

This paper seems to suggest that immigration of non Anglo-Saxons into the USA would increase economic development so the paper must be wrong.

It seems to suggest the opposite, assuming you mean that immigration of non-anglo-saxons would increase genetic diversity.

I see an early front runner for next year's IGNOBLE awards.

I just read the paper. It's overwhelming; 100 pages long. I'm not surprised that the editors in Science decided to cover it in their Editor’s Choice section. This is one of the most profound papers that I have read in economics. It’s a real pity that some of the comments that were posted are of such a low quality. Why don't people read at least the introduction of the paper before rushing to make idiotic remarks?

As an anthropologist and archaeologist, Daniel, I have just finished reading this entire article and this is not profound research. This article reeks of the days when we measured heads and racially profiled populations. This article is marked by a deep misunderstanding of the archaeological data it used to support its argument and has frightening conclusions. Full rebuttal coming soon.

I hope you don't propose that the index of expected heterozygous is racially biased? Do you? The paper uses data from the Human Genome Diversity Project and NOT archeological data. Is this index racially biased? Perhaps we should revert to the story of creation? I thought we are dealing with science, not theology.

Daniel, No I am not. I am arguing that their misinterpretation of this genome diversity project data is flawed. Their assumption that I quote on pg. 3 of the article "according to anthropologists, groups are historically native to their current geographic locations and have been isolated from genetic flows" is completely wrong. Please look at the list of populations listed on the project website, these include Han Chinese (North and South): You've got to be kidding me to assume that they were isolated from genetic flow. Maya, French.... I could construct a rebuttal argument to each one of these. Its also interesting that they give NO citations whatsoever to support this claim......

This paper does also claim to use archaeological data, however they only cite Diamond's 1997 article, they misread his argument as unicasual, and have not bothered to look at ANY of the other literature with regards to why agriculture arose at different times around the world. They choose to ignore that local ecological conditions played a huge role in the development of agriculture and hence populations densities and provide no argument to the contrary, other than to say... "Oh but we think genetic diversity, which is correlated" played a more important role here". Correlation is not causation and a much more careful examination of the evidence is needed before one can much such broadly sweeping statements.

Replying to this statement: Jason Malloy
“Using a large data set of skull measurements and an analytical framework equivalent to that used for genetic data, we show that the loss in genetic diversity has been mirrored by a loss in phenotypic variability. We find evidence for an African origin, placed somewhere in the central/southern part of the continent, which harbours the highest intra-population diversity in phenotypic measurements.”

Archaeologists and anthropologists have argued against this kind of faulty reasoning for years!!!! This statement is akin to something which would have been published in our fields at the turn of the century.

As I'm seeing that reading anything published in archaeology beyond Jared Diamond seems to be out to the intellectual reach of the authors of this paper, I'd like to suggest they at least read the popular text, "The Mismeasure of Man" by Stephen Jay Gould, which summarizes why using such measures are inherently wrong and neatly summarizes why we think this is the case.

"This statement is akin to something which would have been published in our fields at the turn of the century."

Well some 'scientists' fancied measuring skulls back then and making up stories about them, but there is still the small issue of an 'analytical framework equivalent to that used for genetic data' not existing at the turn of the century (or more than a few decades ago), certainly a complication when bringing up old criticisms of old methods to disparage this study?

Ken, I am not disparaging the genetic dataset per se, but the erroneous assumptions the authors of this paper make about the relationship between genetics, and factors such altruism, cooperation and distrust. In support of their flawed logical connection, in their appendix H, the authors cite a number of papers in Evolutionary Biology and gives examples of studies on Drosphila, honeybees and birds. There is not a single citation from any paper in Biological Anthropology, many of which have studied the complexity of the factors which go into deciding how humans enact altruism, cooperate or do not cooperate with each other. Please start by reading "The Foraging Spectrum" by Robert Kelly and marvel at how more complex this issue is than the simple genetics.
I do not believe that any of my geneticist colleagues (who work on human genetics) would approve the statements made in this paper.

I think there has been a confusion, the quote from Malloy that you responded to was taken from a different unrelated paper (in order to make a point about the amounts of African genetic/phenotypic diversity), not the original paper about what that might have to do with economic development. I could certainly agree that Appendix H of the original paper is on the speculative side as far as it applies to people!

Jade, as an anthropologist, you must have heard of Luca Cavalli-Sforza. Here is a quotation from one of his articles in Nature on the Human Genome Diversity Project (the data source of the Ashraf-Galor paper): “All five continents are represented in the collection, and all samples are from populations of anthropological interest — that is, those that were in place before the great diasporas [of] the fifteenth and sixteenth centuries, when navigation of the oceans became possible. This choice [of populations] is important, because these diasporas caused significant population admixtures, especially in the Americas but also in other continents. Only genetic knowledge of the original populations that contributed to these admixtures can disentangle the various genetic complexities that resulted, and the HGDP fulfills these criteria.” So, while these populations may not have been 100% immune from admixture, the sample here is pretty much as good as it gets.
As for your rant on the authors' (mis)interpretation of Diamond, if you actually read the paper and understood the empirical exercise in its entirety, you would have seen that the author's ARE accounting for ALL the factors emphasized by Jared Diamond's work in their regression analysis. Which brings me to my next point -- do you actually know what the term "exogenous variation" means? Rather than flailing around cliche's like "correlation does not imply causation," perhaps you should have someone who understands instrumental variables regression analysis read the article first and then explain it to you. The author's are EXTREMELY careful with their analysis in terms of establishing a causal effect statistically. Also, as an aside, Diamond is neither an anthropologist nor an archaeologist. He is a geographer.

Daniel, Lets talk about regression, exogenous variation and how it is used here:

This paper treats the genetic data as each population having an entirely independent history both from a genetic and from a historical point of view, when in fact, they are highly correlated and inextricably entangled with genetic population structure and with contingent historical events.

The following article describes why this is the case (Cavalli-Sforza is a co-author):
J.Z. Li, D.M. Absher, H. Tang, A.M. Southwick, A.M. Casto, S. Ramachandran, H.M. Cann, G.S. Barsh, M. Feldman, L.L. Cavalli-Sforza, R.M. Myers. Worldwide human relationships inferred from genome-wide patterns of variation. Science. 319(5866), 1100-4 (2008).

You need to assume that your data is independent to carry out this kind of analysis and it is not.

I'm also arguing that their study is fundamentally flawed because it assumes that there is a causal relationship between genetic diversity and complex behaviors such as innovation and distrust. If you look carefully at their tables you will notice that some of the genetically most diverse groups (San) are Hunter-gatherers: societies for which cooperation and sharing are essential mechanisms for survival. Which brings me to another point. Ops. It looks like the authors forgot that we WERE hunter-gathers for 99% of human history, regardless of genetic background, yet their argument predicts that these patterns should be present since populations moved out of Africa.

Accounting for all of Diamond's factors does not solve the problem when their first and most basic assumption is flawed.

Now here is a question for you economists: Is this article actually considered to be important scholarly work in your field or has it already (and rightly) been shut down?

Here is our initial response to Ashraf and Galor's paper:

I just came across Ashraf and Galor's response to Guedes et al. -

As a cultural anthropologist, I am proud of their response that encourages interdisciplinary dialogue on this fascinating topic.

Comments for this post are closed