Louis Putterman responds to my reading club post on Putterman-Weil.  Reprinted with his kind permission:

I just wanted to offer one correction and one further quick comment.

You wrote: “How could one even begin to construct such a matrix?  Whenever possible, P&W use actual genetic data, then supplements genetics with history.” For better or worse, this isn’t really the case. We rely heavily on heroic assumptions that unless identified by sources like the ones mentioned in our appendix such as Everyculture.com, Countriesquest.com, Encyclopaedia Britannica, World Christian Encyclopedia, etc., people living in a country conventionally assumed to be fairly ethnically homogeneous, e.g. France or Spain, are descended from people of that same country. The sources do mention migrants, for instance Algerians in France, but there may well be an undercount of mobility of the presumptively French population, who could well have ancestors who crossed what is now a border with Germany, Italy, Spain, or even a couple of borders (from, say, Poland) sometime in the 500 years ending in 2000. (We have to hope that such migration doesn’t make a big difference, as descendants will in the meantime have perhaps become for all practical purposes like those with only ancestors within current French boundaries during that half millennium.) We used sources based on genetic studies only to get estimates of the regional ancestries of populations described as being of mixed origin, e.g. “mestizos”, in those same sources, and only for countries having a large share of such people, 30% or more, for instance Mexico. The method is described on p. 1632 of our paper, and I think I can see how your misunderstanding might have arisen based on the first sentence on that page: “whenever possible we have used genetic evidence as the basis for dividing the ancestry of modern mixed groups that account for large fractions of their country’s population.” As you can see from a closer reading, we do this only for mixed groups such as “mestizo” “mulatto” etc. and only when they are a large fraction. The extant DNA studies mainly attempt to pick up differences between long separated populations, such as sub-Saharan Africans and Europeans, but not between members of populations that haven’t been as separated, like Germans and Italians.

My other comment is that you write that “civilized migration – where people voluntarily move to a new country to peacefully improve their lives – is an extreme historical rarity.” We don’t take any definite position on that, but off hand it seems wrong and we didn’t mean to suggest it. Although there was a lot of forced movement of Africans to the New World early on, there was overall even more voluntary movement to the Americas, mostly of people from Europe but also ones from other regions, and likewise large parts of the overall migration to places like Australia, New Zealand, South Africa, Singapore, and Taiwan was voluntary on the parts of the migrant.  Whether these arrivals were welcomed by the original inhabitants (Native Americans, indigenous Australians and Taiwan indigenous peoples, etc.) is a whole other story.

By the way, regarding state history working well with 5% discounting, I believe we tested a bit and found it relatively insensitive, so we used the discount that had been applied in other papers. It does turn out that if one lacks back to years before 1 CE, some results become more sensitive to discounting. Borcan, Olsson and I have a working paper in which we report state history for roughly the same number of countries, but going back to the first states, in Mesopotamia before 3,000 BCE. We find that current GDP is concave in this longer-term state history especially when using a low discount like 1% so that the more ancient periods still get non-negligible weight; by concave, I mean that the oldest states such as Iraq are predicted to have lower GDP than “middle aged” ones like England and Germany.