Surnames, mobility, and assortative mating

Those are the topics of a new paper by Güell, Mora, and Telmer, which is interesting on multiple levels.  The abstract is here:

We propose a new methodology for measuring intergenerational mobility in economic well-being. Our method is based on the joint distribution of surnames and economic outcomes. It circumvents the need for intergenerational panel data, a long-standing stumbling block for understanding mobility. It does so by using cross-sectional data alongside a calibrated structural model in order to recover the traditional intergenerational elasticity measures. Our main idea is simple. If ‘inheritance’ is important for economic outcomes, then rare surnames should predict economic outcomes in the cross-section. This is because rare surnames are indicative of familial linkages. If the number of rare surnames is small this approach will not work. However, rare surnames are abundant in the highly-skewed nature of surname distributions from most Western societies. We develop a model that articulates this idea and shows that the more important is inheritance, the more informative will be surnames. This result is robust to a variety of different assumptions about fertility and mating. We apply our method using the 2001 census from Catalonia, a large region of Spain. We use educational attainment as a proxy for overall economic well-being. A calibration exercise results in an estimate of the intergenerational correlation of educational attainment of 0.60. We also find evidence suggesting that mobility has decreased among the different generations of the 20th century. A complementary analysis based on sibling correlations confirms our results and provides a robustness check on our method. Our model and our data allow us to examine one possible explanation for the observed decrease in mobility. We find that the degree of assortative mating has increased over time. Overall, we argue that our method has promise because it can tap the vast mines of census data that are available in a heretofore unexploited manner.

There are ungated versions here.  For the pointer I thank the excellent Kevin Lewis.


Here's my review of Gregory Clark's 2014 surname analysis book, "The Son Also Rise:"

Clark did a fairly good job of persuading me that these low social mobility rates aren't just an artifact of social-climbing name changers, but I would like to see more analysis directly measuring the effect that a small percentage of the ambitious change their surnames to higher status ones each generation. For example, the family story is that the Sailers used to be the Seilers until an ancestor got himself picked mayor of a small town and changed the spelling to something higher class than plain ropemaker.

There is something to be said for this theory, very popular in the Third World (people assuming the names of royalty, notorious people from history, etc). And most generations die out. In fact, unless you have five or more offspring, the chances are your DNA will, over time, die out. For your name not to die out you need at five sons or more. In fact, there's a theory, genetic drift is one part of it (, that everybody over a very long time will assume the characteristics of the most dominant DNA person on earth, which I think is Genghis Khan. Khaaaan!!!!

Having the surname "Kennedy" seems to be somewhat advantageous in political circles.

Not as good as 'Bush' or 'Clinton' these days.

But better than Biden (thank goodness).

My wife once observed that there's hardly a surname from the British Isles that doesn't have alternative spellings. Does the problem arise in Catalonia? How should such a problem be dealt with?

I was thinking this too. It probably makes the data analysis more sturdy for going back a few generations, but considerably less meaningful for an analysis stretching back several hundred years (based on my experience with US genealogical research). Basically, when society was less literate, or when widespread literacy was still recent, the spellings of names was more fluid (coincidentally, so were things like exact birth years) even among those who were literate. So while different spellings is a good indication of un-relatedness within say 3 generations, it's an exceedingly poor indication of un-relatedness within say 10 generations. I have a huge number of ancestors around the 1700s and early 1800s whose names were mostly spelled in a way different from their own children, so my sense is that this was still an active issue in the US at that time. I don't know just when it died down in different European countries, but my guess would be mostly before the 20th century.

Somebody who wanted to analyze 20th century inheritance patterns might be in a better spot for all the different spellings of certain names. But somebody who wanted to do an analysis like Piketty, some sort of dynastic analysis over many centuries, might have to assume that all the weird name variants are the same - even though this seriously taints the data.

Any such dynastic analysis would probably have to get informed about naming origins. Some surnames look similar but may have very different origins, which is the case with my surname which is only one letter different from a much more common surname. They are unrelated genealogically, so they should never be lumped together as a proxy for inheritance. Any centuries-long census data analysis would need to gather information on the different variants of the selected uncommon surnames and filter out spellings used by multiple national origins.

I don't know if this is really true, but I've heard that in Scandinavia if your name is Lars Thorson, your first son will be named Thor Larson. You'll alternate names. All your sons will have the last name Larson and your daughters will have the name Larsdottir. If that's true, genealogy must be pretty tough in Scandinavia, even though they keep meticulous records.

I think we now have the technology, with Big Data and gene sequencing, to map the ancestry of everyone on earth with some level of accuracy. The cost may be prohibitive now, but won't be in the future. We would no longer have to use surnames as a proxy for family lines, but could identify the actual family lines.

If we had full genomes, we could also identify genetic markers that are correlated with upward mobility. This wouldn't happen in the United States due to the political climate, but could happen elsewhere.

Good comment. All these questions are going to be settled in a few decades. We're going to have a fairly complete genetic family tree to ask questions of.

"If ‘inheritance’ is important for economic outcomes, then rare surnames should predict economic outcomes in the cross-section."

tl;dr. Are they defining "inheritance" very broadly? Because you get a lot more from your parents besides cash.

We yes and no, we could determine much of the world's ancestry, but we'd have problems with a number of cases e.g.:
1. Identical twins share genomes and thus we will have trouble figuring out who descended from which twin. Likely a small factor, but potentially significant for specific individuals.
2. Adoption. When looking at generational mobility and inherited social capital, adopted parents are not going to show up. This could be quite useful for measuring the impact of genetics vs social capital, but most of us have an adopted ancestor in the last ten generations.
3. Cuckolding. Similarly, adultery likely exists that will quickly make a hash of inherited political capital. For instance Richard III shows that there was most likely some adultery in the English royal line. Just because your genetic great[n]grandfather was an archer instead of the king doesn't mean that your great[n-1]grandfather wasn't born with massively more social capital than normal.
4. It is possible to have ancestors who contribute nothing to your DNA. While half of your DNA comes from each parent, the distribution between grandparents is more random. Normally, this is a small factor, but extrapolating back enough generations (non-materilineal/patrilineal) will eventually give rise to ancestors who contributed no unique genetic information. Particularly for populations that have low variation, you will likely reach a point where potential ancestors become indistinguishable.

Specifically, the set of all ancestors of anyone alive today, and the set of people who are common ancestors of everyone alive today, in any exogamous population, converged much more rapidly than intuition would suggest. Almost everyone in a community or nation is descended to some extent from everyone in a modest founding population within a few centuries, in the absence of extremely powerful caste or geographic barriers. Even the relative proportion of ancestry in modern individuals from each of the members of the founding population homogenizes to a shocking extent in that time frame.

#4 When talking about surname and DNA, it is about the yDNA only along the paternal line. However yDNA does mutate so there is a margin of uncertainty.

Iceland has both outstanding genealogical coverage of its isolated population and an aggressive genomics firm.

There is another set of obvious and powerful reasons for a decline in mobility.

Europe had this thing called World War II. Deaths and serious injuries of workers in the war, interruptions of careers to adapt to the war, massive war related economic disruption to settled fortunes and business ventures, and wholesale regime change in almost every economy in Europe resulted. World War II created countless opportunities and had a major leveling effect. Seventy years (a bit more than two generations) later, the dust created by World War II has settled and people who were successful in the first or second generations preceding the current one have had time to consolidate their gains during a largely prosperous and peaceful era and to try to pass them on to their children, sometimes with success.

Also, the post-war period opened the door to higher education and big business positions to meritorious members of entire social classes that had been denied it for centuries. In the first generation, this is disruptive and leads to massive social mobility. In the second and third generations, it creates a national economic elite chosen on the basis of merit rather than privilege to a greater extent that any European economic elite for centuries. This genuinely meritorious elite was unsurprisingly better able to hold onto its advantages and pass it to the next generation than the rotten hereditary economic elite in place in Europe at the dawn of WWII.

Given that there are some genetic components to the talents and abilities which contribute to success in a meritocracy (though luck and many other factors are involved as well) and combine with assortive mating and the "ratchet" that successful parents can often cushion their less driven or talented progeny from reverting to the mean. This would suggest that over time, a culture that removes barriers to lower class success will first become less unequal, as the talented from the lower classes rise and the upper classes moderate a bit. But then, as this shift completes, the culture should move toward more inequality, as there would be fewer and fewer talented lower-class people to take advantage of the old mechanisms and the strivers of yesterday consolidate their gains. Of course, there will always be outliers, but overall this is what such a model would predict. The wild card is immigration/emigration which can either drain the ambitious or add a new lower-class pool. This simply restarts the cycle, as most immigrants come in at the low end of the social scale, but given reasonably free societies, rise to their "natural" level within a generation or three. One might expect immigration to increase inequality in the short run, decrease it in the medium term, and then increase it again still later.

I agree that the predictions about more (and more rigid) inequality follow from this model. One thing you said doesn't make sense: How could emigration skim off the ambitious, while immigration adds only a lower-class pool? Don't the talented and ambitious emigrants immigrate somewhere? For every brain drain, there is a brain geyser somewhere else, no?

For me, another important wildcard is class-specific fertility. Greg Clark found evidence that the most fertile class in 18th century England was the upper class, whose kids gradually replaced the lower classes. This kind of pattern seems unlikely to return, but a society that pulls it off could end up looking quite different from one where the opposite happens.

Another wildcard is the prospect of targeted embryo selection.

It depends on which group is emigrating to where. Easier journeys might not select as heavily for the most ingenious/determined. And few people emigrate to nations where the standard of living is lower than where they are, so immigrants tend to be lower-class when they arrive at least. A middle-class Indian is going to be pretty poor in America, whereas the poorest American would be middle class in India. One can see this in the differential in status between, say Mexican immigrants and Chinese. Some Chinese immigrants are fairly affluent already these days, but still many come to work in restaurants and the like. Mexicans don't have to buy a plane ticket, the barriers are lower, so the model would suggest a more mixed bag of immigrants.

Comments for this post are closed