A series of figurative light bulbs went off in economist Gregory Clark’s mind after a reporter asked him a question about names a half decade ago.
Flip: everyone’s name carries information about them, in particular aspects of their social status.
Flip: people’s interest in their own heritage means we’re creating big databases containing, among other things, listings of names stretching back well into the past.
And flip: watching those surnames ebb and flow over times can tell us something about social mobility.
And so begins David Edmonds’ conversation with Clark, a Scottish-born economist who teaches at the University of California, Davis. And while Clark’s research interests include economic growth over the long-run and the wealth of nations (in particular England and India), his illuminating foray into viewing social mobility through the prism of surnames has resulted in one book, The Son Also Rises: Surnames and the History of Social Mobility, and a lot of headlines.
Here, Clark discusses the surname research that has gained him such notoriety, the genetic bases for social mobility, and whether or not government intervention can change the playing field or if instead surnames are destiny.
A spoiler alert: for most populations –excepting say, the Copts in Egypt—social mobility reverts to the mean, so the mighty are eventually humbled and vice versa. But this, Clark, suggests, is a natural phenomenon, and not an induced one.
Looking at two European examples, England and Sweden, Clark finds data that upsets the conventional wisdom: “[T]here’s been no change in social mobility rates between medieval and modern England. But then, the move was to Sweden because that we expect to be a society of very high rates of social mobility. There’s enormous government interventions in Sweden designed to ensure social mobility. Startling information from Sweden is again: the rates are no higher than in modern England in modern Sweden. And in 18th century Sweden, the rates are as high as they are in modern Sweden.”
Just maybe, he hypothesizes, there’s a “physics of social mobility.”
Nigel Warburton: What’s in a name? Well, according to Gregory Clark, quite a lot. Surnames predict social position with remarkable accuracy. Social mobility follows universal patterns and, surprisingly, only usually occurs very very gradually over a number of generations. Why this is so is an interesting question; as is the question of whether this is something we should worry about.
David Edmonds: Greg Clark, welcome to Social Science Bites.
Greg Clark: Thank you. Great to be here.
DE: The topic we’re talking about today is “Names”. That’s an odd topic. How did you get into that?
GC: It was a complete accident. It was actually a question from a reporter for the New York Times. If you’d have asked me six years ago would I write anything in my life about social mobility or about names, I’d have said no. But it turns out that names are this amazing link that everyone carries that’s linking them – in some cases – to thirty generations in the past.
DE: So was it a kind of light bulb moment when you realised there was this huge mine of data that you could access?
GC: It actually turned out to be a series of light bulbs. First of all, you realise the concept that if names carry information about social status, the rate at which that information is lost from the social system is a measure of the entropy of the system and of the rate of social mobility. And that that measure could be applied all across different societies. And then the second realisation was that the intense interest in people about their ancestors had meant that in the last few years, we’ve arrived at this juncture where giant databases are available revealing the distribution of surnames, and the distribution of surnames amongst elites, in many societies, over many centuries. And then, the third kind-of light bulb was to realise that all you needed to measure the rate of social mobility was information over time on what’s the general distribution of surnames and what’s the distribution amongst some elite. That, with that, you can derive complete measures of social mobility rates.
DE: So you study different countries and also you’ve looked at different time periods?
GC: Yes. I started with England; discovered surprisingly slow rates of social mobility. But then, everyone has the impression that England is hidebound, moribund, locked in the past. And so, the first thing then was to go back to medieval England. And then, the startling discovery is that there’s been no change in social mobility rates between medieval and modern England. But then, the move was to Sweden because that we expect to be a society of very high rates of social mobility. There’s enormous government interventions in Sweden designed to ensure social mobility. Startling information from Sweden is again: the rates are no higher than in modern England in modern Sweden. And in eighteenth century Sweden, the rates are as high as they are in modern Sweden. And so that actually then started a kind of hypothesis in my mind which is: well, maybe, social mobility rates are in fact a social constant; that there’s a physics of social mobility. And that led to then investigating a whole bunch of other countries.
DE: So that was your hypothesis. Was it vindicated by what’ve found?
GC: So far, yes. And the most extreme cases here were communist China where, in 1949, you had a revolution which sought to overturn the social order; which executed large numbers of the previous ruling class; where large numbers of members of that class fled to Taiwan, or to Hong Kong or to the United States. And where, using surnames, what you can observe is that there’s maybe been a slight increase in social mobility rates but the leadership of the communist party now disproportionately represents the elites under the Qing emperors.
DE: This is an astonishing finding because, as you say, everybody assumes that the UK is very class bound and compared to the social-democratic Sweden and communist China, you would expect there to be much less social mobility perhaps in the UK than those countries. But that’s not what you’ve found.
GC: No. The very strong surprise here is the constancy of this rate of social mobility. And one nice example we can use from Britain is Oxford and Cambridge. We can look at names that show up disproportionally in these universities in 1800 and see how quickly they have become more average. And so, nineteenth century Oxford and Cambridge are gentlemen’s clubs. You have to have Latin to get in; you have to go to particular elite schools; they have their own ceremonies and exams; you have to attend Church of England service. If you move to Oxford and Cambridge now, these institutions have transformed themselves. They now use publicly available exams that are available in every school in the country. They rely much less on interviews; they rely more on standardized measures. Astonishingly, they have not – in one iota – increased the rate of social mobility. Elite names that were there last generation are still much more likely to be there this generation. And one reflection of this is if you have rare English surname, and someone with that surname was at Oxford or Cambridge around about 1800, you are now four times more likely than the average person to attend Oxford or Cambridge. On the basis of that one piece of information.
DE: So how many generations does this effect last for?
GC: One of the things I should emphasize is we do find universal tendency for status to regress to the mean. That’s actually a comforting feature of society; no elite in any of our studies remains an elite forever. And no underclass remains an underclass forever. How long getting to the average takes depends how far away from the average you are when you start. In the case of England, we can observe some groups that are so elite that it takes twenty five generations for them to become average. And, from the bottom end, it turns out it’s harder to distinguish the really underclass groups in societies. But, even then, sometimes you’re talking ten generations for a group to go from below-average to average. I should mention there’s one exception to this rule which is that we can actually find some societies where elites don’t obey this law. But that’s associated with marital endogamy within the group. So, for example, the Copts in Egypt since the Muslim invasions have been an elite and that’s for 1500 years and haven’t shown much sign of progressing to the mean. And if you go to India, we can find almost complete stability of some elite groups. But again, associated with this lack of endogamy. So, I should qualify this by saying that in any society wherein marriage is not completely endogamous amongst elite groups, there will be regression to the mean.
DE: Can I ask you about the methodology of this? Presumably, some countries have better data sets than others. There must be all sorts of problems comparing different generations. How do you go about resolving those kinds of issues?
GC: Yes. Very much, in every country, we had to grapple with unknown problems of data interpretation. What we need for this method to work reliably is an unerring rule that you inherit the surname of your father. And there have been many deviations. Swedes, for example, have very little loyalty to their surnames. And large scale name-changing is actually occurring in Sweden. Mostly, that’s out of low-status surnames towards higher-status surnames. But fortunately, in Sweden, the authorities maintain a register of every name and you can’t change your name without legal permission from the tax authority. And, we can actually do nice name-control then and make sure that we are using names that are actually inherited. In a society like Britain, one of the traditional freedoms of the British is the ability to change their name any day they want and to change it as often as they want. But, interestingly, in Britain people have much more name fidelity. There are many unfortunate British names now – Sidebottom, Broadbottom, Nutter – that people still hold onto. And they’re not actually even associated particularly with low status. Another one is: Glasscock, Mycock, Hercock, Hiscock; people hold on to these names. And, for our purpose, that’s actually a big advantage. But there is some selected name-changing and we have to deal with that. And then, another problem is to have access to information on name distribution; some societies like Japan have a very great suspicion of public information. That makes it very hard. We need to know what the general name distribution is, what it is amongst elites. England, increasingly, is moving to a society where it may become impossible to do this type of research. Because, now even to acknowledge that someone attended Oxford and had a particular name is not permitted under British information laws.
DE: So social mobility does take place but over many generations. What are the mechanisms by which people revert to the mean?
GC: For me, as a social scientist, this is the most interesting part of the story. The question that really comes up is: is this a cultural phenomena? Is it a familial culture that is being passed on? Is it resources? Or is the basic genetics of inheritance? And if it’s cultural or resources, what it says is that societies are dramatically failing to achieve appropriate rates of social mobility; that it’s a number one problem for all societies; that President Obama is right to say that this is the problem his administration will tackle. If it’s just a basic issue of genetics and of assortative mating and then the transmission of certain types of abilities or competencies, then actually two things: one is we don’t have a problem. And the second thing is we shouldn’t devote enormous resources to trying to deal with it.
DE: So let me press you on that. So, one interpretation is that all that’s happening is that intelligent people are passing their genes on to intelligent people and so remaining in the elite. Another is that there are barriers to a meritocratic society. Which is it?
GC: My own personal bet is that genetics plays a much greater role in this than people have been willing to consider. One test would be, in cultural explanations, your grandparents; your cousins; your other relatives should all have some influence on your outcomes. If you’re from the Jewish community, for example, then being part of that larger community network should have significant influence on your outcomes. In a genetic interpretation, if we truly knew the status of your parents – the underlying status – that would be the only predictor of your outcomes. Your grandparents, all the rest of the stuff would not matter. And also, things like resource shocks should be relatively unimportant. And interestingly, again using Oxford and Cambridge data, we can actually test. do your parents only matter or does your more extended lineage matter to predicting future success? And the answer is: it’s only your parents. If we can get the data, it’s only your parents that matter. Another test is a genetic explanation would say that any elite group that only intermarry internally would not actually regress to the mean. Because, the genetic information is not being lost from that group. And that, again, we can test by looking at various examples. And what we find then is that in societies with a high degree of endogamy, the rate of social mobility doesn’t seem to slow down. Another interpretation here would be that any elite group would simply have been selected from a larger population by some mechanism – or any underclass group. And we again can test this by looking at history and saying for example: Ashkenazi Jews are elite; Sephardic Jews are elite. Are they a subset of a larger population? And the answer overwhelmingly and very clearly is yes. Only a small fraction of the original Jewish population has survived as Jewish. The rest converted to Christianity. And, there’s very strong evidence that that was the elite share of the population. And, we can also see in modern America that new social elites are actually being formed by immigration policy which means that people coming from areas distant from the US, without familial connections to the US, are being drawn from very high-level elites in those societies. So now, the super-elites in the US are Coptic Christians; Indian Hindus; Iranian Muslims; Maronites. And when you look across those groups, what you see is – culturally – an incredible diversity. The only groups that are not represented now in the modern US elites are protestant Anglo-Saxons. (Laughs.) So what you actually see when you look at this data, it seems to me, is that eliteness has nothing to do inherently with culture; it’s to do with the familial transmission of abilities.
DE: Is that one reason why your research, which has received a lot of publicity, has been given a rather pessimistic spin?
GC: Yes, absolutely. And it’s puzzling to me – having thought about this for some time – why there’s a universal assumption that the good society would have to have a high rate of social mobility. And, secondly, that it’s kind of a dismal discovery that genetics could actually predict what people’s outcomes are going to be. Because it seems to me that a truly meritocratic support with universal support for everyone, in the end the only thing left that would explain differences would be genetics. And so the book, in some sense, is saying you’re much closer to the good society than you thought; you only thing you didn’t expect was the high degree of predictability of people’s outcomes based on their lineage.
DE: So people have been rather depressed by your results. But you’re saying they should be comforted?
GC: Yes! I actually take it as a sign that we live in a surprisingly fair and meritocratic world, even in many early societies
DE: If your interpretation of your own findings is correct, I guess the biggest policy implication might be that governments around the world are wasting their time and money investing in trying to make society more meritocratic.
GC: I have to be careful to say that there’s not evidence in the book that a whole bunch of individual programmes that governments are attempting are not worthwhile. But the evidence is that that would make a very small difference to the overall rate of social mobility; that the dream of a society where every generation is born anew is a hopeless endeavour. Instead, what we have to think about is in a society where there is strong predictability from lineage, how do we want to organize that society in response to that finding, in terms of how much inequality do we want to allow in outcomes. And I take this empirical evidence as support for a very – kind-of – Rawlsian interpretation of social order. That our fate in life is largely determined by lineage and, in a world like that, we have to be modest about our own contributions to these outcomes and realize that it would be very wrong to excessively reward those who have the right lineage and penalize those who don’t. That’s something where we know societies have been able to have enormously different outcomes. Sweden versus the United States; that Sweden is a much better society on that basis. And now what we’re learning is that, given the realities of social mobility, even more is it going to be the case that the Swedish vision of social order is more appealing than the American one, which relies on the idea of universal and rapid social mobility.
DE: Do you see it as part of your role to come to policy implications and to make normative judgments about your findings? Or are you just crunching the numbers?
GC: I’m mainly interested in the description of the social world, and in some sense, in the physics of the social world. But, of course, one of the interesting things about doing social sciences is that it’s so closely related to our thinking about how we should order the world that we live in.
DE: It sounds like you see yourself more as a scientist than a social scientist.
GC: One of the disappointing things about doing economics is how little we’ve been able to do that has any real flavour of science. Economics is an enormously ideologically dominated subject. We have almost no laws that are either not trivial or instantly falsified. One of the astonishing things here was to discover that underlying what seems to be a very complex social world, there’s a very simple social physics. And a physics that allows me, in cases like England, to make predictions two hundred years into the future about what the relative social status of groups will be. And so, actually I have drawn great comfort from the idea that even in economics we will sometimes find powerful and simple regularities that make interesting and surprising predictions. Testable predictions, as well. So, actually I draw a lot of comfort from that. And I think, in the end, there will be no distinction between physical science and social science and that this is a sign that really these things are much more closely intergroupal than we thought possible.
DE: Is the idea of testability, of falsifiability, of predictability crucial to what you do in the robustness of your claims?
GC: Absolutely. We’re not saying anything until we’re making predictions about the social world. Part of the research, in fact, conducts tests which say we don’t know yet between people and their great-great grandfathers. With this model, what would we predict? And now we’re collecting data where we can go in and test and say: how close did we get, in terms of the model? And so, I absolutely buy into the idea that to do any systematic, scientific investigation of society is to be able to make predictions.
DE: Greg Clark, thank you very much.
GC: Thank you. This has been very fun.