As I noted in a postscript to my last post, Quamrul Ashraf and Oded Galor have prepared a response [Update: the response is no longer online] to the Harvard academic critique of their paper on genetic diversity and economic development (I recommend having a look through the comments on that post, where Jade d’Alpoim Guedes, Nick Patterson (both authors of the critique), Henry Harpending and others continue the debate).

Apart from the broader question of whether this work should even be undertaken, the Harvard critique focused on two issues: causation and the statistical foundations of the work. Ashraf and Galor are quick to dismiss the statistical critique:

[O]ur critics have falsely suggested that we treat socioeconomic and genetic data as if populations are independent of one another. On the contrary, our empirical analysis accounts for the possibility of spatial dependence across observations, including analytical methods that correct for spatial autocorrelation in “error terms” and bootstrapping. This criticism of our work thus reflects either a misunderstanding of the techniques that we employ or a superficial reading of our work.

The response on causation is more detailed, and one of Ashraf and Galor’s arguments is one that I did not expect to see. They write:

The key is that the measure of intra-population genetic diversity that we employ should be interpreted as a proxy (i.e., a correlated summary measure) for diversity amongst individuals in a myriad of observable and unobservable personal traits that may be physiological, behavioral, socially-constructed, or otherwise. ...

A careful reading of our research should make it apparent that our use of the measure of genetic diversity from the field of population genetics does not imply that our hypothesis is one of biological determinism, nor does it imply that DNA material is directly important for economic outcomes or that some genes are more important than others for economic success. The fact that the measure of genetic diversity we use is based on variation across individuals in non-protein- coding regions of the genome (and, thus, in genomic characteristics that are not necessarily phenotypically expressed so as to be subject to the forces of natural selection) is clear reason why our findings should be interpreted through the lens of our measure serving as a proxy for diversity more broadly defined.

The more relevant question to ask therefore is to what extent the measure we use can reasonably be considered a proxy for diversity in unobserved phenotypic or socially-constructed characteristics. There is indeed an emerging body of scientific evidence that establishes remarkable correlations in this regard.

Ashraf and Galor are also quoted running this line in a Nature News piece on their paper:

 Galor and Ashraf told _Nature_ that, far from claiming that genetic diversity directly influences economic development, they are using it as a proxy for immeasurable cultural, historical and biological factors that influence economies.

After reading this, I went back to the paper to confirm my previous understanding of it, and if Ashraf and Galor intended to use genetic diversity as a proxy, it is not clear. The paper appears to finger genetic diversity and the phenotypic expression of that diversity as the relevant causal factors, with no suggestion it is a proxy. For example, they write:

The hypothesized channels through which genetic diversity affects aggregate productivity follow naturally from separate well-established mechanisms in the field of evolutionary biology and from experimental evidence from scientific studies on organisms that display a relatively high degree of social behavior in nature (e.g., living in task-directed hierarchical societies and engaging in cooperative rearing of offspring). The benefits of genetic diversity, for instance, are highlighted in the Darwinian theory of evolution by natural selection, according to which diversity, by permitting the forces of natural selection to operate over a wider spectrum of traits, increases the adaptability and, hence, the survivability of a population to changing environmental conditions. On the other hand, to the extent that genetic diversity is associated with a lower average degree of relatedness amongst individuals in a population, kin selection theory, which emphasizes that cooperation amongst genetically related individuals can indeed be collectively beneficial as it ultimately facilitates the propagation of shared genes to the next generation, is suggestive of the hypothesized mechanism through which diversity confers costs on aggregate productivity.

I would like to see a more direct defence of their argument about the causal mechanisms. However, Ashraf and Galor do suggest in their response that further research on the causal mechanisms is required.

The timing of this debate has highlighted the extent of continued disciplinary divides. Ashraf and Galor released the working paper a couple of years ago, and they have since presented it in a raft of conferences and seminars. It was then accepted for publication in the American Economic Review, but the current debate was only triggered when the paper was mentioned in Science (gated). The pre-publication of working papers so prevalent in economics, and which is starting to gain traction in other fields, still relies on the working paper getting in front of people who might be interested in commenting. The reality is, however, that publication in a reputable journal remains the point at which a paper comes to others' attention - or becomes “important” enough that it deserves a response.

My posts on Ashraf and Galor’s paper on genetic diversity and economic growth are as follows:

  1. A summary of the paper methodology and findings

  2. Does genetic diversity increase innovation?

  3. Does genetic diversity increase conflict?

  4. Is genetic diversity a proxy for phenotypic diversity?

  5. Is population density a good measure of technological progress?

  6. What are the policy implications of the effects of genetic diversity on economic development?

  7. Should this paper have been published?

Other debate on this paper can also be found hereherehere and here.