The Son Also Rises Read online

  From 1920 to 1939, local authorities provided financial support that enabled less affluent students to attend university. After World War II, there was a major increase in government financial support for secondary and university education. Oxford and Cambridge eventually devised entry procedures that should have reduced the admissions advantages for students from the traditional, endowed feeder schools. These measures would seem to imply a much faster regression to the mean for elite surname frequencies at Oxford and Cambridge after 1950. Yet there is no evidence of such a trend in figures 5.7 and 5.8.

  Political Status

  An even more elite group than the wealth holders or Oxford and Cambridge students is members of Parliament. There were about five hundred MPs on average from England and Wales in the nineteenth century, rising to around 550 in the twentieth century.

  Records of members of Parliament extend back to 1295 and thus allow us to observe the mobility implied by movements in and out of the English political elite since the high Middle Ages. Here we take the list of rare surnames found at Oxford and Cambridge for the years 1800–1829. As a social elite, this group of surnames is also overrepresented in Parliament. Figure 5.9 shows the relative representation of these rare surnames in Parliament in the years 1830–2012. The Parliament measure starts in 1830 so that the sample corresponds to the birth cohort associated with entry to university between 1800 and 1829.

  The relative representation of these rare surnames in Parliament follows the same pattern as their association with wealth and Oxford and Cambridge attendance. It again shows a relatively steady decline over generations, though the process becomes noisier at the end because of the small numbers of MPs and the many chance factors that determine whether any particular person becomes an MP. But five generations later, in the 1980–2009 cohort, these rare surnames are still more than three times as frequent in Parliament than in the general population. The implied intergenerational correlation of status is high, at 0.8.10

  FIGURE 5.9. Relative representation of Oxford and Cambridge rare surnames from 1800–1829 among members of Parliament, 1830–2012.

  The remarkable thing about this finding is that over this period, significant changes in political institutions brought new parties and new groups into the House of Commons. Why did these changes not have a greater effect on the representation of surnames among MPs? The allocation of seats in Parliament for counties and boroughs was largely unchanged from the medieval pattern as late as the election of 1831; it failed to reflect the growth and redistribution of the population. Medieval boroughs, such as Dunwich in Suffolk, were entitled to elect two MPs. Over time, with changes in trade and industry, many of these once-thriving boroughs became depopulated villages or hamlets with tiny electorates. Most of Dunwich, in fact, fell into the sea.

  These fossilized borough constituencies became known as rotten boroughs. The landowners of these boroughs effectively controlled the election of MPs. Voting was public in these years, and the small number of electors could be bribed or intimidated to vote as directed. The most notorious example was Old Sarum, which in 1831 had a population of seven and an electorate of four. In the 1831 election, just before reform, 152 out of 406 MPs, more than a third, were elected by fewer than one hundred voters. Thus a large part of the House of Commons was in the control of the traditional landed classes.

  The Reform Act of 1832 disenfranchised 56 such boroughs in England and Wales and reduced the representation of another 31 to only one MP. It also created 67 new constituencies and extended and regularized the franchise. In 1867, a second Reform Act disenfranchised a further two rotten boroughs and reduced a further 35 boroughs to one MP. Ancillary legislation in 1867–68 eliminated another 9 rotten boroughs. The 1867 act also resulted in another major extension of the franchise (though it was still restricted to men). Yet these reforms had no perceptible effects on the rate of intergenerational mobility among the political class.

  Along with these reforms, over time the electoral franchise was extended to men (and then women) lower on the social scale. The franchise extended to only 13 percent of men in 1830 but to 100 percent by 1918. The extension of the franchise was associated with the rise of the Labour Party as the voice of the urban working class. By 1923 the party had 191 MPs, including members from Scotland. From then on large numbers of Labour MPs were returned at every election.

  But, as figure 5.9 shows, the rise of a working-class party in the early twentieth century was not associated with a rapid replacement of traditional elite surnames in Parliament. Whatever the political arrangements, this surname group maintained its overrepresentation in Parliament and in the halls of Westminster. Lineage dominated ideology and party.

  Conclusions

  The tracking of rare surnames shows that social mobility rates for wealth, education, and political power are low in modern England. The intergenerational correlation of status in these various dimensions is on the order of 0.73–0.80 and has not decreased since the nineteenth century, despite the advent of modern economic growth, mass public education, the extension of the political franchise, and the welfare state. Modern rates of social mobility also represent only very slight increases on mobility rates in the medieval period, when for artisan surnames the estimated persistence rates for education and wealth were on the order of 0.80–0.85. These rates are similar to those found in Sweden for recent occupational and status mobility and in the United States for recent occupational mobility.

  The intergenerational correlation of wealth in England is estimated at 0.73 from an analysis of surname groups. Yet conventional estimates of intergenerational wealth persistence, which look at individual families, are always in the range 0.41–0.48. How do we reconcile these different estimates of social mobility rates? Which of these estimates is correct? In the next chapter we offer a theory as to why surname estimates consistently reveal much lower social mobility rates than the conventional estimates do, and why surname estimates are the better guide to overall rates of social mobility and to the underlying structure of status inheritance in societies.

  1 Smith is also a very common surname among the Traveller population, which sends few students to Oxford or Cambridge.

  2 Watson and Galton 1875.

  3 “I’m So Broke” 2013.

  4 As noted, it has ever been the right of the English to call themselves whatever they want, as long as there is no intention to deceive. This ability to change surnames is something we have to guard against in the examples below. But surprisingly few English people have made use of this right, even when their birth name has unfortunate connotations. Shakespeare, for example, when he wants a comic surname for one of his “rude mechanicals” in A Midsummer Night’s Dream (1594), chooses Bottom. But there were still 549 Bottoms in England in 1881, as many proportionately as in the marriage records of 1594.

  5 In each case, wealth was measured as a logarithm to limit the effects of outliers of extreme wealth on the results. See Clark and Cummins 2013 for details.

  6 This database was assembled from a variety of sources, such as exam results published for Oxford from 1976 to 2009 and the e-mail directories of Oxford and Cambridge for 2010–12.

  7 This is calculated defining average representation of a surname as a representation at Oxford and Cambridge no more than 10 percent above its share in the general population.

  8 Winstanley 1940, 83.

  9 Greenstein 1994, 47.

  10 We assume in this calculation that MPs represent the top 0.1 percent of the status distribution.

  SIX

  A Law of Social Mobility

  Even a fool learns something once it hits him.

  Homer, The Iliad

  THIS BOOK ESTIMATES SOCIAL MOBILITY RATES by measuring the rate at which surnames that originally had high or low social status lose that status connotation. If a surname such as Pepys or Brudenell-Bruce had a high status in 1800, how rapidly does that surname regress to average status? If Baskerville was an elite name in the Domesday Book o
f 1086 in England, is there any echo of that distinction in 1300, 1500, or now? The book examines how surnames reflect the rate of social entropy, the rate at which original status information leaves the social system.

  The four previous chapters reveal not one but many surprising results from using surnames as measures of mobility. The first is that in all the cases examined, social mobility measured from surnames is much lower than from conventional measures. The conventional measure, as discussed in appendix 1, just looks at the correlation between parent and child for any aspect of status. Surname status shows regression to the mean in all cases, but the process is slow. Elite surnames can take ten or fifteen generations (300–450 years) to become average in status.

  The second surprising result is that social mobility seems to occur at a similar rate for different measures of status: wealth, education, occupational status, and membership in political elites. Wealth would seem to be much more heritable than education or occupational status. You can leave your indolent or idiot son or daughter a pile of money, but that money won’t get them into medical school without superior MCAT scores. Why, then, does wealth show the same rate of persistence as these other aspects of status?

  The third surprise is that the rate of persistence is close to constant across wildly different social systems. It is little higher for the feudal England of the Middle Ages than for the progressive, equality-loving, social-democratic Sweden of today.

  If individual family estimates of social mobility suggest intergenerational correlations of 0.15–0.60 for these measures of social status, why do surname studies suggest correlations of 0.75–0.80? Why are the results so different when we examine surname groups rather than individual families?

  A Simple Theory of Social Mobility

  The theory proposed here to explain the above findings is simple, but it has significant implications for estimates of the rate and nature of social mobility. The proposal is that we must distinguish between a family’s surface or apparent social status and their deeper social competence, which is never observed directly.1 What is observed for families is their attainment on various partial indicators of social status: earnings, wealth, occupation, education, residence, health, and longevity. Each of these derives from underlying status, but with a random component.

  Thus if yt, for example, measures the earnings of a family in generation t, this assumption just formally translates into the statement

  yt = xt + ut

  where xt is the family’s underlying social competence, and ut is the random component.

  The random component of aspects of social status exists for two reasons. First, there is an element of luck in the status attained by individuals. With respect to earnings, high-earning people happen to choose a successful field to work in or a successful firm to work for. They go to work for Facebook rather than Myspace. They just succeed in being admitted to Harvard, as opposed to just failing. They marry a supportive spouse instead of ending up shackled to a needy partner. Second, people trade income and wealth for other aspects of status. Someone might choose a career as a philosophy professor as opposed to a lower-status but more lucrative career selling plumbing hardware.

  The second assumption in this simple theory of all social mobility is that underlying social status in families regresses only slowly toward the mean, with a persistence rate, b, of 0.75. And this high rate of persistence is constant across all societies. Formally,

  xt + 1 = bxt + et,

  where et is a second random component.2 This is the social law of motion that is tested in the rest of this book. The claim of this book is that these two assumptions are sufficient to describe social mobility in all societies. This insight leads to powerful predictions about social mobility and its sources, predictions that are successfully tested below.

  The above implies, for example, that the conventional studies of social mobility, based on estimating the intergenerational correlation β in the relationship

  yt + 1 = βyt + vt

  for various partial measures of status—earnings, wealth, education, occupation, and so on—underestimate the true intergenerational correlation b that links underlying social status across generations. In particular, the expected value of conventional estimates, β, is not the underlying b but instead θb, where θ is less than one. Further, the greater the random components of any measured aspect of status, the smaller will θ be.3

  Figure 6.1 shows the structure being proposed. The determining variable here is underlying social competence. Because the correlation of earnings with this variable is less than one, the correlation of earnings over one generation is bρ2, where ρ is the correlation of earnings with underlying social competence.

  Since we have these two measures—b for underlying social mobility and β for partial measures of status—why is it that the underlying b is the true rate of social mobility? The reason is that if we were to measure families’ status by an average of the various observed aspects of status, , then

  where indicates an average of the various random components. But as we average status across many aspects—earnings, wealth, residence, education, occupation, health, longevity—the average error component shrinks toward zero. Thus the intergenerational persistence of average social status approximates to b as opposed to β.4 The underlying b gives us a better measure of the persistence of status on average for families, as opposed to the persistence of any particular aspects of status. Thus b best measures the true rate of social mobility.

  FIGURE 6.1. The intergenerational transmission of status.

  Conventional estimates of social mobility, based as they are on estimating the correlation of parents and children on partial measures of social status, systematically overestimate the underlying mobility rate. It is the underlying mobility rate that determines overall social mobility rates for families and social mobility over multiple generations.

  What causes the conventional measures to overestimate underlying social mobility rates is the presence of the error term linking partial measures of status with underlying competence. By looking at groups of people (as long as they are grouped by identifiers that do not correlate with this error term, such as race, religion, national origin, or even common surnames), we can reduce this error term by averaging across the group. While

  yi = xi + ui

  at the individual level, at the group level,

  Now the accurately tracks without the intrusion of the errors, and we can correctly estimate social mobility. When we look at such groups of individuals, the underlying, slow rate of social mobility becomes apparent even when we can observe only the usual partial indicators of underlying social competence. This is why the surname groups provide a measure of underlying rates of social mobility. But any grouping that is independent of the current random elements determining a partial measure of status will do the same. That is why it will always seem that racial, ethnic, and other minorities within societies experience slower than expected social mobility.

  For this argument to hold, it must be the case that the various manifestations of social status are all only loosely correlated with underlying social competence. This loose correlation can be shown both by illustrations and systematically. As an illustration, the State of California conveniently makes available to the public the salaries of all faculty in the University of California system. This information reveals that professors who would be regarded as equivalent by such criteria as level of education and occupational status in fact earn vastly different amounts. Figure 6.2 shows median salaries in 2012 for some disciplines. Professors of English and music earn about one-third the salaries of professors of management. If we were to infer status based only on earnings, we would conclude that there was a vast social gap between these species of academic.

  FIGURE 6.2. Median professors’ salaries by subject, University of California, Davis, 2012.

  TABLE 6.1. Correlations between different aspects of status

  Table 6.1 confirms that any partial measure o
f social status, such as earnings, wealth, education, or occupation is only a very indirect measure of the underlying social competence of individuals. The table shows the average reported correlation for the same individual for several aspects of status: cognitive ability (typically represented by IQ), occupational status, education, earnings, and wealth. The correlations for any two attributes average 0.43. That means, for example, that if we know the cognitive ability (the IQ) of a child, we can typically predict less than one-fifth of the variation in the child’s possible educational achievement, occupational status, earnings, and wealth.5 This loose association of the various aspects of status for any person means that each aspect of status must also be only weakly correlated across generations.

  This simple switch in thinking about the mechanism of social mobility can explain many of the puzzles noted in the existing literature on social mobility. It also enables us to make quite strong predictions about the nature of social mobility. Here are the predictions of this model.

  The observed rates of regression to the mean for individual aspects of status are determined by how well they are predicted by the underlying status of families. The lower their correlation with this underlying status, the lower their intergenerational correlation. Thus mobility rates appear to differ for different aspects of status, depending on how closely each is linked to underlying status. And measured mobility rates vary across societies, again depending on how closely measures such as earnings correlate with underlying social status.

  In the long run, all aspects of status regress to the mean at the same rate. Underlying mobility, as measured through earnings, wealth, education, or occupational status, will be the same.