The Son Also Rises Read online

  FIGURE 5.2. Decay at the family seat: regression to the mean among the Brudenell-Bruces.

  Most of these surnames, however, are too obscure to signal any particular social status: Bigge, Buttanshaw, Hilhouse, Skipwith, Taddy, and Willyams, for example. A typical foreign name from this sample is Bazalgette, discussed in chapter 1. All of the fifty-seven Bazalgettes in England in 2002 seem to be descended from one man, Jean-Louis Bazalgette, a Huguenot immigrant from southern France, who became tailor to the prince regent, later George IV. The prince must have had quite extravagant tastes in clothes, because Jean-Louis left the enormous fortune of £250,000 at his death in 1830. Jean-Louis had two wives and many children. Within a generation, his family had become thoroughly English, with at least seven of his grandchildren serving in the British armed forces. His grandson Joseph William was well known in his own right as the designer of the London sewer system (see figure 5.3). But for all that, the name Bazalgette is unfamiliar to most people.

  FIGURE 5.3. London monument to Sir Joseph Bazalgette.

  The second sample of surnames from the probate register is drawn from families in the top 5 to 15 percent of the wealth distribution. This comprises seventy-six similarly diverse and mostly obscure surnames. The notable ones are De Grey, Pepys, Pigou, and Rothschild. But, as before, most of these names are largely unknown: Brandram, Brettingham, Brideoake, Broadmead, Broderip, and so on.

  The third sample is 237 rare surnames held by poor people. The first source of such names is a list published by the government in 1861 of habitual paupers, people who had been on poor relief (public assistance) continuously over the last five years. The surnames used here are a subset of that list. No one with these surnames dying between 1858 and 1887 evidenced any wealth at all. The names themselves signal nothing of their low status, and indeed some are surprising. The list includes Defoe, the surname of the famous writer Daniel Defoe (1660–1731). Though Defoe is a French-sounding surname, it is very rare in modern France, and Defoe himself was born Daniel Foe, adding the De- as a pseudo-aristocratic affectation.4 Because Daniel Defoe had surviving sons, and there is no record of any Defoe in English births or marriages before his birth, the modern English Defoes likely mostly descend from Daniel. The poverty of the Defoes who died between 1858 and 1887 reflects downward mobility.

  The Inheritance of Wealth in Modern England

  The people identified in these initial rare-surname groups were born, on average, around 1813. How did their descendants fare in the four to six generations up to the present?

  As noted, the probate records give an indicator of wealth at death for every adult dying in England and Wales from 1858 onward. A variable fraction of adults was probated at death in each year: only 15 percent in 1858 but 42 percent by 2011. Of the estates probated, all have an assessed value. For wills not probated, we have no measure of the value of the estate. But since most of those avoided probate because their assets were of insufficient value, we can assume for this group an average estate value of half the minimum value of those requiring probate. In this way we can assign an average probate value to all rare surnames by generation of death, starting in the period 1858–87.

  Figure 5.4 shows the result. Wealth for each surname group in each period is measured relative to the wealth of the average adult at death. The initial wealth of the rich surname group is 187 times that of the average, and for the prosperous group, it is 21 times the average. Both rich and prosperous surname groups regress toward the mean with each generation, but by 1999–2012, four generations later, they are both still much wealthier than the average person at death: the rich are four times as wealthy, the prosperous three times as wealthy. The initial wealth of the poor group is half the average. By 1999–2012 this figure was an estimated 90 percent of the average. Figure 5.4 demonstrates that even among the great-great-grandchildren of the Victorian generation, wealth disparities between surname groups persist.

  Table 5.2 shows the implied intergenerational correlation for wealth in each generation for each group of surnames.5 Because many of these surnames are so rare, we can identify individual fathers and sons for more than four thousand cases. Thus the first column of the table shows the estimated intergenerational correlation for individual families using conventional measures to estimate the persistence rate. This persistence rate, ranging from 0.41 to 0.48, is in line with the other, limited evidence on wealth inheritance within English families. Notice also that there is no sign of any increase in wealth mobility between the generation dying in the years 1888–1917 and that dying in the years 1999–2012, despite the many institutional changes in the intervening years.

  FIGURE 5.4. Wealth at death relative to the average, by surname type.

  TABLE 5.2. Intergenerational correlations for wealth at death

  The other columns show the intergenerational correlations for average wealth across rich, prosperous, and poor surname groups. For all three groups, the correlation averaged over the entire period is much higher than that revealed by conventional measures: for the rich it is 0.71, for the prosperous 0.77, and for the poor 0.64. The estimate for the poor, however, is one potentially contaminated with random error, as can be seen by the large fluctuations in estimates for different generations. So the estimates for the rich and prosperous are the focus here. For these groups, again, we see no sign of any increase in wealth mobility between 1888 and 2012. Yet the earlier generation held wealth in an era when taxation and redistribution of income and wealth were very modest compared to more recent periods.

  The maximum inheritance tax rate in England for those dying between 1858 and 1887 was 4 percent. Thus this generation could pass on wealth almost intact to their heirs. In contrast, for those dying between 1960 and 1993, the maximum inheritance tax averaged 69 percent (see figure 5.5). The wealth inherited by the generation dying between 1999 and 2012 thus would have been heavily taxed, and this taxation should have pushed their wealth much more quickly toward the mean than was the case for previous generations. Yet no such trend is apparent in the data. The persistence of wealth remained just as high for the last two, heavily taxed, generations as for the previous two, which escaped significant inheritance taxation.

  Under the various inheritance-tax laws in effect in England and Wales from 1858 to 2012, considerable wealth could be exempted from tax. Since allowances sometimes varied for unmarried and married testators, it is not possible to estimate the exact tax burden in each period. But for members of the rich surname group dying between 1960 and 1993, of all wealth bequeathed, roughly 57 percent was absorbed in taxes. It is thus surprising that the persistence of wealth in the fifth generation, the beneficiaries of these bequests, is just as strong as in previous generations.

  Other evidence of the strong persistence of wealth differences by surname comes from probate rates in each surname group. Since typically only the wills of people with wealth at death were probated, and a large fraction of the population has no wealth at death, these rates are another good indicator of surname-group wealth.

  Figure 5.6 shows the probate rates for the three surname groups at death over five generations. Wills for those in the rich and the prosperous surname groups are currently probated at a rate of 60 percent, compared to 45 percent for the general population. Probate rates for the poor are converging toward that of the average surname, but only in the fifth generation, and even that convergence is not yet complete.

  FIGURE 5.5. Maximum inheritance tax rates, United Kingdom, 1825–2012.

  FIGURE 5.6. Probate rates for surname groups, by generation.

  From these probate rates, we can infer the intergenerational persistence of wealth in England, using the method given in appendix 2. Table 5.3 shows the estimated correlations for the rich, prosperous, and poor groups by generation and averaged over generations. These confirm the estimates above based on actual wealth estimates.

  These wealth measures have drawbacks as a general index of social mobility. First, it may be objected that of various
components of social status—education, occupation, earnings, health, and wealth—the most persistent is wealth, since it can be directly inherited. Second, the measures of wealth discussed above may not fully reflect the social changes that took place in Britain in the twentieth century. For the last generation we observe, those dying in the years 1999–2012, the average date of birth was 1924. These people would, on average, have completed their schooling by 1946, before many of the social changes of the postwar era. This raises the question of whether social mobility might be much greater for people in England born more recently.

  Greater longevity is making the circulation of wealth in modern economies increasingly socially dysfunctional. In 1858–69, when our data on deaths begins, the average age for those whose wills were proved was 62. Given an average gap of thirty years between generations, wealth was inherited by children on average at age 32, just as they were rearing their own children and buying housing. But now people are, on average, fifty years old when they inherit any wealth from their parents. By then they typically own their own house and cars, and their own children have completed much of their education. If longevity continues to increase, then, despite increases in the average age at which women produce children, wealth will increasingly pass from the ancient to the aged.

  TABLE 5.3. Wealth correlations from probate rates

  Inheritance of Educational Status

  As a way of measuring social mobility in current generations, we can turn to education. Specifically, we look at the fraction of people admitted to Oxford and Cambridge as an index of educational status. This measure can be applied to people born as late as 1994.

  From 1830 to 2012 these universities typically enrolled only 0.5 to 1.3 percent of each cohort of the eligible domestic population. The data used here provide a complete record of Oxford and Cambridge attendees from 1800 to 1893, and thereafter four-fifths or more of attendees through 2012—a total of six hundred thousand students. The last birth cohort observed is those born in 1993–94.6

  For this purpose, because the surname sample is small and admission to the universities is a low-frequency event, the rare surnames of the rich and the prosperous dying between 1858 and 1887 are combined into one group. Figure 5.7 shows the relative representation of these surnames at Oxford and Cambridge by thirty-year generations starting in 1830, ending with a current cohort of entrants, 2010–12. The figure shows how overrepresented these surnames are at Oxford and Cambridge relative to their incidence in the general population.

  The rare wealthy surnames show up at very high rates in the initial generation of university entrants, 1830–59. Someone with one of these surnames was fifty times more likely to enroll at Oxford or Cambridge than someone from the general population. Over the next six generations, the last observed being university entrants from 2010 to 2012, there is a substantial decline in the relative representation of these surnames. By the last generation, a member of this surname group was only six times as likely to enter Oxford or Cambridge as someone in the general population. This finding implies that simply by knowing that people with a given surname born circa 1813 on average died wealthy, it is predicted that people born circa 1990 with the same surname are six times more likely than the average person to attend Oxford or Cambridge. This is a remarkable persistence of status over six generations.

  There are thirteen observations of the wealthy rare surnames at Oxford and Cambridge for the period 2010–12, compared to an expected two. This high number could be in part just the result of luck. Admission to Oxford and Cambridge depends on exam results, secondary school location, and sympathetic interviewers, among other factors. However, the probability that this high representation between 2010 and 2012 is all due to random chance can be shown to be one in a thousand. Rare surnames that were elite in the early nineteenth century remain elite today.

  FIGURE 5.7. Relative representation of rare wealthy surnames at Oxford and Cambridge, 1830–2011.

  Even so, the admissions data show a decline in relative representation for this surname group. What does this trend imply about the intergenerational correlation of educational status in England? The figure also shows the pattern of decline that would be predicted if the correlation of educational status across generations were always 0.82. The fit is generally good, despite the small numbers of observations in later generations.

  Educational status persists for this group even more strongly than wealth. Moreover, as with wealth mobility, there is no sign that educational mobility has increased in the past few generations. An intergenerational correlation of 0.82 fits the pattern reasonably well across all generations. The implied rate of mobility is so low that the rich elite surnames will not approach an average representation at Oxford and Cambridge for another seventeen generations (510 years).7

  The rare surnames in this sample are all associated with wealth in the nineteenth century. Could their persistence as an educational elite be due to the association of these surnames with wealth? Have other families in the educational elite of 1830–59 regressed to the mean much more quickly?

  To test this hypothesis, we draw from the university records another, larger rare-surname group that consists of any surnames with five hundred or fewer holders in the 1881 census that appear on the lists of entrants to Oxford and Cambridge between 1800 and 1829. Thus all we know about these surnames, apart from their rarity, is that they show up on the university rolls sometime in this period. In 1830, this group of surnames represented just over 1 percent of the population of England but nearly 12 percent of all Oxford and Cambridge students. Figure 5.8 shows the relative representation of this much larger surname group in subsequent generations at Oxford and Cambridge.

  Again the high educational status associated with these surnames erodes very slowly. They are still overrepresented in the current student population, although that overrepresentation is now modest. Such surnames are only 65 percent more frequent in the universities as in the population of all eighteen-year-olds. But again, this finding implies a strong intergenerational correlation of status over six generations. The persistence rate that best fits this pattern is 0.73.

  FIGURE 5.8. Relative representation of Oxford and Cambridge rare surnames from 1800–1829 among Oxford and Cambridge students, 1830–2012.

  The pattern of declining relative representation predicted by such a level of persistence in educational status is shown in the figure: it fits the observed pattern of decline very well across almost all the generations. The implied intergenerational correlation of status seems constant between 1830 and 2012. Later generations show no increase in social mobility. The only deviation is that in the years 2010–12, an unexpectedly high number of the elite surnames from the period 1800–1829 are observed. But this anomaly may be due to the fact that our data for current students comes from a different source (the university e-mail directories).

  Thus intergenerational persistence is just as high for education as for wealth. And in this case, the persistence extends to the generation born 1992–1994. Both the wealthy and the educationally privileged of 1800–1829 are losing their elite status only slowly. Yet since that time the nature of universities and the way in which they recruit students have changed dramatically.

  The finding about the persistence of rarer surnames at Oxford and Cambridge is remarkable. Suppose you are now eighteen and have a surname that was held by five hundred or fewer people in 1881. Even if all we know about you is that someone with your surname was enrolled at Oxford or Cambridge in the years 1800–1829, then we can predict that you have a 65 percent better chance of being admitted to one of those universities now. If you have a surname that was even rarer in 1881, held by forty or fewer people, we can improve on that prediction: if a person with your surname attended Oxford or Cambridge between 1800 and 1829, then your chances of admission are 3.5 times better than those of the average eighteen-year-old in England.

  These findings hold despite significant changes in admissions policies and financial s
upport for students in the intervening years. In the early nineteenth century, Oxford and Cambridge were largely closed to those outside the established Church of England. Not until 1871 were all religious tests for graduation finally removed. As late as 1859, one member of our wealthy surname group, Alfred de Rothschild, who was Jewish, had to petition to be excused from Anglican services at Trinity College, Cambridge. This exemption was granted as a special indulgence.8

  Before 1902 there was little or no public support for university education. Oxford and Cambridge offered scholarships, but most of them went to students from elite endowed schools, who were coached to excel at the scholarship exams. In the years 1900–1913, nine schools, identified as the elite of English secondary education in the Clarendon report of 1864 (including Eton, Harrow, and Rugby), supplied 28 percent of male entrants to Oxford.9 In addition, entrance requirements favored students from more exclusive educational backgrounds: before 1940, entrants to Oxford, regardless of their intended field of study, were required to complete a Latin entrance exam.