Social mobility, fixed forever? Gregory Clark’s The Son Also Rises is a book of scholarship, and of scholastic overreach

[ This post is a book review of: Gregory Clark (with Neil Cummins, Yu Hao, and Daniel Diaz Vidal and others), 2014. The Son Also Rises: Surnames and the History of Social Mobility. Princeton and Oxford: Princeton University Press. ]

The Son Also Rises forcefully advances the idea that social position is determined by innate inherited abilities, an idea that is potentially pregnant with policy implications. “Once you have selected your mate,” Gregory Clark counsels, “your work is largely done. You can safely neglect your offspring, confident that the innate talents you secured for them will shine through regardless.”

With this book Professor Clark (an economic historian with the University of California at Davis) dons the mantle of Francis Galton, who more than 100 years ago examined the transmission of status across the generations of 19th century England, and who is equally known for the statistical methods he developed to study the issue.

Gregory Clark, an economic historian, used statistical methods, originally developed by Francis Galton, in his historical study of surnames and social mobility.
Francis Galton, a Victorian scientist, developed a method called “regression to the mean” as a part of his research on biostatistics and human genetics.

And in the same way that careful readers have both admired and also questioned Galton’s work, so too they can legitimately admire The Son Also Rises as a work of scholarship, while at the same time wonder about it as an exercise in scholastic overreach.

Social scientists have examined many different indicators of the intergenerational transmission of status. Clark draws conclusions from his and his collaborators’ tracking of the status associated with surnames. Galton used height, but earnings and income are particularly salient in recent research, and occupation, education, wealth, and other indices have also figured prominently. The intergenerational transmission of earnings, for example, varies across countries, with the level of economic development, with the degree of inequality, and in some cases over time as policies and institutions change. Other indicators have been found to show less variation across time and space. Whereas much of this research looks back only one, at best two, generations, Professor Clark makes a major contribution in tracing the transmission of status back centuries.

Figure 1.6 From page 9 of The Son Also Rises
Figure 1.6 from page 9 of The Son Also Rises, showing conventional versus surname estimates of status persistence. “Conventional” refers to an average of intergenerational associations of incomes and education.

The book holds that previous studies are based on observable manifestations of the intergenerational transmission of a latent characteristic, which earnings, education, occupation and all the other indices measure imperfectly. Clark sometimes refers to this underlying characteristic as “social competence,” and at other times outright as “social genotype,” but ultimately what he means is “genes.” When measured correctly, social competence has a high degree of intergenerational transmission, with about three-quarters of any relative advantage or disadvantage being transmitted from parent to child. This equals or exceeds Galton’s estimate for height, and is as much as three times some estimates of the transmission of earnings between fathers and sons.

But more important, according to Clark social mobility is an unchanging constant. It did not change with the arrival of free public education, the fall of nepotism in both the private and public sectors, economic growth, the expansion of the franchise in the 19th century, or even with the rise of the welfare state and redistributive taxation in the 20th. Social mobility is not related to any of these policies, and it is not tied to the level of inequality.

This, he claims, is a law for all times and all places, from 14th century England, to 20th century China, from the America of the settlers to the America of Wall Street financiers.

I shall call it “Clarke’s law”, my deliberate misspelling of the author’s name paying homage to his use of surnames and changes in surnames, and to the use of the letter e to represent influences other than family background—what Clark refers to as “luck”—in the mathematical representation of a son’s status as a linear function of his father’s status. Clarke’s law states that 50% to 80% of variation in status is predictable at birth. There has always been social mobility, albeit very slow, but it has little to do with society, institutions, or, by implication, public policy.

The book’s first part, an exercise in description, lays out the evidence for this law in countries for which surname information is available for the longest period. The approach involves tracking surnames through time, with due care that spellings may change, and linking them to some measure of social position.

For Sweden, this involves associating uncommon surnames among the Swedish nobility and the educated elite of the 17th and 18th century to the social status of those carrying these same surnames today, as indicated by the names appearing in lists of physicians, attorneys, university students, members of the Royal Academy and even among the high earners in publicly available tax records. The names of titled nobles, for example, appear on the Swedish Bar Association’s register of member attorneys at nearly six times their proportion in the general public, and Clark estimates the persistence of high-status occupations, in general, at about 0.75.

Similar methods are applied for the United States and the United Kingdom, using elite names culled, for example, from the American Medical Association’s directory of physicians, lists of licensed attorneys, or the list of federal tax payers published in The New York Times in 1923-24.

The Domesday Book is a “survey and valuation of all the land held by the King and his chief tenants, along with all the resources that went with the land in late 11th century England” Source:

To put these three modern cases in relief, Professor Clark offers surname evidence from Medieval England dating back to 1300. Elite names are identified as people associated with Oxford and Cambridge Universities, with wills proved in the highest court, with those probates associated with “Sir” or “Gentleman,” and with Members of Parliament. The Norman conquerors of England in 1066 recorded as property owners in the Domesday Book of 1086 are 16 times more likely than other names to be represented in Oxford or Cambridge in 1170 and 25% more likely now: a steady, but very slow, regression to the mean, suggesting a persistence rate of 0.90 from one generation to the next. The bottom line is that “medieval England had mobility rates similar to … those of modern United States and Sweden. In terms of social mobility, then, what did the Scientific Revolution, the Enlightenment, and the Industrial Revolution achieve? Very little.” (pages 74 and 75)

This is scholarship at its best, but it also conveys a whiff of overreach that becomes manifest in the second part of the book, which sets out to test the law in diverse countries (India, China, Taiwan, Japan, Korea, and Chile) and also among particular groups (Protestants, Jews, Gypsies, Muslims and Copts).

Claiming that this dynamic applies to society as a whole rather than just to elites, Clark is sensitive to the need for supporting evidence. He repeatedly seeks confirmation that the data are representative of entire populations. Yet at times the analysis is very local—for example, the use of two specific regions in the Shanghai area as the basis for the study of the evolution of status among Chinese surnames.

Figure 3 Panel B Corak and Heisz 1999
These Canadian data for sons born in the 1960s show that the strength of the tie between their adult earnings and the earnings of their fathers varies with paternal earnings, and is strongest for those children born to top 1% fathers. If the view advanced in The Son Also Rises were applied to these data we would expect to see a horizontal line.  Source: Corak and Heisz (1999), “The Intergenerational Earnings and Income Mobility of Canadian Men: Evidence from Longitudinal Income Tax Data,” Journal of Human Resources, figure 3, page 525.

Even if it is debatable whether or not the various data are representative, it may still be appropriate to claim the law is widely applicable—so long as the proportion of advantage passed between generations does not vary across the social hierarchy (that is, the mobility process is linear). But if Clarke’s law varies with social status, then the book is only about the propagation of elite status across the generations.

The research community examining contemporary patterns in social mobility struggles with both of these issues while using much more extensive and more accurate data than can be imagined for past times. That these issues could be resolved more cleanly by even the best economic history is a suggestion that many may find surprising.

Clarke’s law can be described as a simple regression to the mean model in the spirit of Galton, one in which two components (reflecting the inheritance of genes and the impact of what some might call luck) shape a child’s adult status. But other models can lead to equally high persistence of status. Powerful parents may influence child outcomes directly, rather than through inheritance, as a result of social institutions, and things like primogeniture, nepotism, access to select colleges, or lax wealth and estate taxation. The children of the elite will succeed independent of their innate talents because of this type of social tracking.

Or it may be that the host of other influences on child outcomes should not be thought of as a random draw, as simple luck with no multigenerational echo. A longstanding family culture may foster capacities to, for example, teach children to speak with a socially approved accent, buffer them from the downsides of health risks and other threats to their human capital, or simply to instill an identity of entitlement. A surname is not just an index of genes but also of social pressures and entitlements that keep some down and keep others from falling down. In this way, what Clark terms luck may also persist across generations and it will also generate persistence in “social competence.”

Whether or not Clarke’s law is valid or can be generalized, there is, nonetheless, an important lesson for public policy that should not be lost. Economic development, progressive income redistribution, the rise of public education, improvements in health, and the collective insurance of the welfare state are not inconsequential. The suggestion in this book is that they matter less for relative differences among individuals than we might think or hope, but surely they still matter for the absolute level of well-being and the development of human capacity—a message that may be lost by a focus on relative outcomes.

The author suggests his law implies that earnings and other economic outcomes should not be as unequally distributed as they are in some countries, because they reflect the transmission of genes and therefore competencies beyond individual control. But this is no more than a passing suggestion, and readers will suspect that Clarke’s law will be interpreted just as forcefully in some public policy communities as an argument for the just deserts of those on top of the social heap.

Ultimately, as interesting as The Son Also Rises is as an exercise in historical scholarship, it does not advance contemporary public policy. Galton even began an 1892 book (Hereditary Genius: An Inquiry into its Laws and Consequences) with the same focus as Clark’s beginning and ending of The Son Also Rises: “I propose to show … [Galton wrote on page 1] that a man’s natural abilities are derived by inheritance, under exactly the same limitations as are the form and physical features of the whole organic world. Consequently, … it would be quite practicable to produce a highly-gifted race of men by judicious marriages during several consecutive generations.” We have heard this advice many times over the decades, and it serves neither good public policy nor good parenting.


[This is the author’s version of the book review originally published with the title “Social mobility: fixed forever?” It is posted here by permission of the AAAS (American Association for the Advancement of Science) for personal use, not for redistribution. The definitive version was published in Science Vol 344 (#6186), May 23rd, 2014, DOI: 10.1126/science.1251568 .

I gratefully acknowledge the helpful comments and suggestions of Bruce Bradbury, Andrew Cherlin, and Dalton Conley on a first draft of this post, but remain entirely responsible for the final result.]


12 thoughts on “Social mobility, fixed forever? Gregory Clark’s The Son Also Rises is a book of scholarship, and of scholastic overreach

  1. Population genetics is a well developed field (with keen equations, just like econoimcs), so it would make sense for economists to talk to population geneticists before they invoke genetic explanations (or when they review the work of other economistss who do). The genetist could tell the economist how hard it is to sort out environmental and genetic effects, and how implausibly strong the selection for Clark’s hypothesized trait would have to be in order to give the result he claims.

    1. Thank you John. Do you have any specific suggestions for further reading? An important review article or two from the field that could open the gateway to the major messages, and other important research?

  2. I know even less about population genetics than I do about economics. I am not able to comment on specifics of population genetic equations, but when I read a sentence like “once you have selected your mate, your work is largely done” I think by way of analogy: horse racing, breeding and sires. There is perhaps no area in which selective breeding is more rigorously practiced than in horse racing. So how well does mate selection in this rarefied, controlled world predict the “work is largely done”?

    Certainly better than random, yet the predictive power is still rather low and uncertain. Take Northern Dancer, perhaps the most important stud in racing history. Of 635 foals, 488 started in races. 368 were winners, but the quality of competition in the races is unknown. Importantly, 146 were stakes winners (higher quality competition), just over 20%. 26 of these animals were champions in Great Britain, Ireland, France, Italy, The US and Canada. Not bad, and yet only about 4% even remotely approached his level of success (and zero reached the level of their father). And important to remember that Northern Dancer is considered the best stud ever by many and so the numbers are better than average.On the other hand, Secretariat, perhaps the greatest race horse in history, fathered 663 foals of which only 53 went on to become stakes winners (with one, Risen Star, winning the Preakness and Belmont).

    Not factored into the above outcomes (and implied ‘equations’) is the higher quality of training, jockeys and opportunities a horse sired by a Northern Dancer or Secretariat will receive. After all, if one is paying large stud fee, you can be sure they will get the best. Yes, I think one could make the argument that even in the world of horse breeding there are many unfair advantages due to “family background”.

    So if even in horse racing (where the ‘traits of success’ are relatively small and clear) the statement ‘once you have selected your mate, your work is largely done’ does not stand up well to scrutiny, I think there is good reason to support John’s statement “how implausibly strong the selection for Clark’s hypothesized trait would have to be in order to give the result he claims”.

  3. It’s not hard to sort out environmental effects. They’re small. The consensus is settling uncomfortably I might add, against the postmodern preference, that it’s 80% genetics and 20% something else, and we don’t know what something else is yet – and environment doesn’t seem to be it.

  4. Ron Wray makes an important point re “the higher quality of training, jockeys and opportunities a horse sired by a Northern Dancer or Secretariat will receive”. It seems that Clark’s use of unusual surnames introduces a confound, in that people bearing them may likely be assumed to be related to the prominent individual of the same name, and will benefit from favorable treatment by friends, colleagues, business partners, etc.; an effect beyond any contribution of genetics or direct family endowments. Cf. Eliza Doolittle, “The difference between a lady and a flower girl is not how she behaves, but how she is treated.” Hard to see a way to remove that effect. I hope I don’t have to read the book to find out if Clark found a control…

  5. I live in Sweden, and was surprised and shocked by the evidence that Clark mustered to show that social mobility has not changed despite all the efforts we have put into increasing it the last century.

    If it is true that much of one’s adult place in the social hierarchy is determined early on, then the argument for a welfare state, high taxation, etc. are much stronger than if people can rise through hard work or fall through laziness. Incentives just don’t matter that much, and that is a good thing. Let’s make sure everyone has a pretty good life :).

    He does consider your objection – that this is all about persistence of social status among the elite – at location 702 in my Kindle edition, toward the end of the Swedish section. He argues that, if your argument is true, once members of an elite family have fallen out of the elite, then that group should rapidly regress to the mean. In that case, there would be a bimodal distribution, with high persistence of status among the elite, and then a normal distribution for the rest. That pattern does not exist.

  6. I’d also note that successful persons have been known to assume surnames associated with elite status. See, for example, “Le veritable D’Artagnan”.

  7. Before I bought the thesis, I would want to know a good deal more about the methodology. Surnames are often passed by adoption, people change their names to higher-status forms (“Smith” to “Smythe”) and so on. Most of these effects would dilute the strength of Clark’s findings.

    1. Actually, he and his co-authors have suggested that they’ve carefully addressed these kinds of concerns, and the book spends a certain number of pages examining the rate at which names change and how the researchers addressed the issue. So they are aware of the problem, and have tried to correct for it.

  8. Couple of points:
    1) Suppose the results were the opposite and your outcome had no correlation with your parents or ancestors then we ought to ask what type of society would that be? Is this a society where nurturing counts for nothing? What are the consequences of this? Would we want to live in such a society?

    2) The question of social mobility starts to become redundant when we reduce inequality between people, not just in terms of wealth, but access to fundamental things like health or education (or opportunity say – whatever that precisely means). In such a place correlations may continue to exist but this may be for behavioural reasons e.g. I spent my life in academia, so did some of my children due to their exposure to me and so did some of theirs. Assuming it’s never prosperous to work in academia, mobility, by looking at wealth at death, would be similar to a society where inequality is deep and you did what your family did because there was no choice. Of-course it’s plain to see the two are different because in the first case my children were afforded the opportunity to pursue other paths.

    So here is question for the author, Clark or anybody interested in this subject. It may be that society does not require that mobility happens at high rates, rather that it is real in the minds of it’s people. Is this enough to shift societal behaviour e.g. crime, life expectancy, education, trust, love?

  9. I feel like you are ignoring the adoption studies and Georgia land lottery experiment he cites. Adoption studies show that people adopted to families with high socioeconomic status are more similar to their biological than their adoptive parents socioeconomic status when they reach adulthood. This is true even if they have been adopted at birth.

    Cant see why the most obvious and plausible explanation for this would not be genetics.

What do you think?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.