The Son Also Rises Read online

  This kinship strategy was pursued in various ways. Kin groups allotted a portion of land as “land for education” (学田), with the rent being used to pay for education and exam preparation. These funds supported lineage-based schools (族学) in which the most academically talented children within the kin group were coached for the exams by the best available teachers. To diversify risk, other children were assigned to alternative occupations.

  Success in the exams not only brought glory to the common ancestor of the chosen candidates but also secured protection for the property rights of the entire kin group. As it obtained more land and wealth, the group had more resources to invest in education. Thus the pooling of resources made the status of kin groups more stable over generations than that of individual families.21 On this interpretation, it is the kin network that explains the strong persistence of some family surnames among the Qing elite as well as elites in Republican and Communist China.

  However, the status persistence observed in China is no greater, either now or in the past, than in highly individualistic societies such as Sweden, England, or the United States. In these Western societies, aid from kin networks is limited and rare. Thus kinship networks are not an essential factor in accounting for slow social mobility. The law of mobility presented in chapter 6 assumes nuclear families with no collateral transfers. It is more parsimonious, then, to assume that China has simply experienced the same strong underlying transmission of social competence within families that we see in the rest of the world, and that kinship associations did little to further reduce rates of social mobility.

  1 The Civil War of 1642–49 saw only a temporary overthrow of the traditional elite, and they returned in full force with the Restoration of 1660.

  2 Schaffer 2012, 2.

  3 Stavis 1978, 67.

  4 Moïse 1983, 142; Stavis 1978, 75.

  5 Stavis 1978, 75.

  6 The irony is that Mao himself was the son of a rich farmer and so himself a “class enemy.”

  7 Guo and Min, 2008; Gong, Leigh, and Meng 2010. Wu and Treiman (2007), however, argue that taking into account urban-rural inequality, the intergenerational correlation of incomes would be significantly higher.

  8 Supposedly these surnames appeared at a similar frequency even during the Song dynasty, 960–1279 CE (Liu et al. 2012, 342).

  9 It is assumed that the relative frequency of surnames was the same in the period 1645–1905 as in 2010.

  10 These surnames represent, however, only 0.055 percent of the modern population.

  11 Based on the numbers listed per year, there were more jinshi than high Nationalist officials.

  12 Of these soldiers, 12,737 were from South Jiangsu and 8,907 from North Zhejiang.

  13 This assumption seems reasonable because there is evidence that common surnames in the lower Yangzi did not change in frequency between 1645 and 2010. And the common surnames are found at the same frequency among dead soldiers as in other local sources.

  14 Hao 2013.

  15 Hao 2013, chapter 2.

  16 Hao 2013, chapter 3.

  17 Walder and Hu 2009.

  18 Ho 1964.

  19 Campbell and Lee 2010, 2011. In a recent paper, Avner Greif, Murat Iyigun, and Diego Sasson argue that these kin networks were important enough in preindustrial China to explain why Chinese technological development was much slower than European (Greif, Iyigun, and Sasson 2012).

  20 Elman 1992.

  21 Hymes 1986.

  TEN

  Japan and Korea

  Social Homogeneity and Mobility

  CONVENTIONAL STUDIES SUGGEST that modern Japan is a socially mobile and meritocratic society. In this view, although Japan before the Meiji restoration of 1868 was a society of rigid class divisions, the reforms of 1868 and 1947 transformed it into an egalitarian, homogeneous, and classless society. Does a society such as Japan, with a high degree of cultural homogeneity, have faster social mobility than societies such as the United States, which are fractured by religious, racial, and ethnic differences? Korea is a society of similar cultural homogeneity, profoundly remade after World War II as a supposedly modern and meritocratic society, so the same questions arise there also.

  Japan

  The largest traditional elite in Japan was the samurai, the former warrior class (figure 10.1). By 1868 they had evolved mostly into bureaucrats and administrators. Because by then they constituted 5 percent of the population, their economic circumstances were diverse, but on average they remained an elite. With the Meiji restoration, the samurai lost the legal privileges they had enjoyed under the shogunate. The new government did, however, compensate them for their hereditary land revenues with government bonds. In 1871 the government ordered all samurai to surrender their swords, made commoners free to intermarry with samurai, and abolished restrictions on the occupations samurai were allowed to pursue. A new education system was launched in 1872, premised on selection for higher-level institutions through competitive examination. The samurai thus soon lost all class privileges in the rapidly modernizing Japanese economy.

  FIGURE 10.1. Samurai of the Satsuma clan during the Boshin War period (1868–69).

  As part of its modernization program, the Meiji government sought to enforce a shared cultural identity and set of cultural practices. The reformed educational system promoted a standard form of spoken Japanese nationwide. The government was aided in this effort by the relative homogeneity of language, culture, and even physiognomy among most Japanese.

  After the Meiji restoration, the new leadership, as part of its Westernization program, merged the kuge, the ancient court nobility of Kyoto, with the daimyo, the feudal lords, into an expanded aristocratic class. The resulting kazoku peerage initially consisted of just 427 families. However, the Meiji government expanded the hereditary peerage by adding to its ranks persons who had made distinguished contributions to the nation. As table 10.1 shows, the total membership grew most rapidly between 1884 and 1907. Thus the kazoku families represented an elite of wealth and position in Japan that dates mainly from before 1907, though new families were being added even after 1928.

  The kazoku had a number of privileges in addition to any private wealth that they retained from pre-Meiji times. Some received hereditary pensions from the state. The titles and pensions passed by inheritance to the eldest son. Only the holder of a title was considered part of the kazoku: other children had no special status. The kazoku elected representatives from their ranks to serve in the House of Peers.

  TABLE 10.1. Number and rank of kazoku families, 1884–1946

  The new constitution of 1947 eliminated the imperial elite: the kazoku was abolished, and these families were all rendered ordinary citizens. Thus for the last two generations Japan has been a society of exceptional social homogeneity. Visible racial and ethnic minorities are few: they include the Ainu, some islanders, and the descendants of Korean immigrants from the colonial era. Social policy has even managed to render invisible the burakumin, the descendants of the outcast communities of the feudal era whose occupations rendered them defiled and impure. Religious minorities, such as Christians, have remained a small share of the population. And in general the Japanese now register in the World Values Survey as the most secular population in the world.1

  Sociological studies support the idea that Japan is an intergenerationally mobile society. One study, for example, looked at the occupations of three cross sections of fathers and sons in 1965, 1975, and 1985.2 If we assign an income level to each of these occupations, we can translate these results into the intergenerational earnings correlation. Using average earnings by occupation gives an implied persistence rate of 0.3 in all these cohorts.3 A more recent study of intergenerational income mobility found implied intergenerational correlations of 0.3–0.46.4 Thus, on conventional measures, Japan has plenty of intergenerational mobility.

  TABLE 10.2. Class status of university graduates, 1890–1900 (%)

  SOCIAL MOBILITY AMONG THE SAMURAI IN THE EARLY
MEIJI ERA

  As the class of bureaucrats and officials, the former samurai began the Meiji era with considerable advantages. These were reflected in admissions to the newly formed universities and technical colleges, in which the descendants of the samurai class were heavily overrepresented. Table 10.2 shows the percentages of samurai and commoners in the major higher-education institutions in 1890 and 1900. Since the descendants of the samurai constituted 5.3 percent of the population, in 1890 the samurai were graduating at twelve times the expected rate from imperial universities, and commoners at less than half the expected rate.5

  The proportion of commoners in the universities rose quickly, however. By 1900, samurai were graduating at only nine times the expected rate. However, because the samurai were so heavily overrepresented initially, the rate of decline in their advantage is still slow once we calculate the implied persistence rate.

  Table 10.3 shows the implied relative representation of the samurai in the universities and the implied intergenerational correlation of educational status. This analysis assumes that university students represented the top 1 percent of the status distribution and that by 1890 the samurai had the same variance of underlying social status as the general population. The implied intergenerational correlation is high, 0.66–0.72. A wider estimate, based on a more comprehensive set of higher-education programs, suggests an average persistence rate of 0.73 for this period.6 Thus even though the raw data of table 10.2 might appear to suggest high rates of social mobility, it actually implies slow mobility rates during the Meiji era.

  This intergenerational correlation in status is calculated assuming that the samurai share of the university-aged population remained constant between 1890 and 1900. There is reason to believe that, if anything, their population share was declining. For example, the source that gives samurai and commoner populations in 1881 estimates an average family size for samurai of 4.54, compared to 4.78 for commoners.7 Any decline in the share of samurai in the young population between 1890 and 1900 would imply a smaller decline in relative representation of samurai at universities and professional schools and hence an even higher persistence rate.

  TABLE 10.3. Implied persistence rates for the samurai, 1890–1900

  TABLE 10.4. Percentage of samurai government officials and persistence rates, 1872–82

  Year

  Central government

  Local government

  1872

  78

  70

  1876

  78

  —

  1882

  61

  58

  Implied persistence rate

  0.71

  0.72

  Harry Harootunian gives the share of central and local government employees in 1872 to 1882 who were of samurai origin.8 These findings are summarized in table 10.4. The share of samurai among government officials also greatly exceeded their share in the general population in 1872. But as with graduates of the universities, this share began immediately to decline. Again, however, the rate of decline was consistent, with a high rate of persistence of 0.72. (This persistence rate again is calculated assuming that the samurai population share was constant over time.) So despite the many social and economic changes occurring in Japan in the Meiji period, the rate of downward mobility of the former samurai is once again the standard rate. This was another social revolution that had surprisingly little bite on the social elite.

  MODERN SURNAME ELITES

  After 1947, Japanese sources no longer categorize people by samurai or kazoku ancestry. These earlier distinctions of caste and privilege have been washed away in the world of the salaryman. But again it is possible to measure the fates of these earlier elites, and their rate of social mobility, by turning to the rare surnames that were associated with samurai or kazoku ancestry.

  A candidate list of samurai surnames was formed from a genealogy of samurai families assembled by the government in 1812.9 Similarly, a genealogy compiled by descendants of the kazoku allows construction of a complete list of surnames once held by kazoku.

  By 1898 surnames in Japan had become strictly hereditary, with little possibility that the rare surnames of the elite were being adopted by less distinguished families. The 1898 Family Registration Law dictated that each household have a surname inherited by children, with a married woman adopting her husband’s surname.10 Adopted children took the surname of the head of the family.11 The Family Registration Act of 1947 established that only the head of a family could apply for a surname change; if granted, it applied to the entire family. Surname changes were to be granted only in cases of “unavoidable reasons.” We thus assume little surname changing after 1947.

  Measuring inheritance of position by surname for Japanese elites is potentially complicated by the prevalence of adoption among upper-class groups. In Japan, where there is no male heir, it has always been quite common for high-status families to adopt a son-in-law to carry on the family name and lineage. Indeed, figures on adult adoptions in Japan in recent decades suggest, remarkably, that as much as 10 percent of each male cohort has been adopted by another family, and that this pattern has prevailed since at least 1955.12 Samurai and kazoku families without sons traditionally engaged in this practice on a large scale.13 A study of the samurai in the Tokugawa era, for example, suggests substantial rates of adult male adoption.14

  Since men will only be adopted as sons if they are at least moderately successful socially, adoptions potentially bias estimated persistence rates upward. However, the bias is likely modest. Suppose, for example, that a quarter of the inheritors of high-status surnames are adopted sons. Suppose also that in these cases, the sons match the status of the adoptive fathers exactly. If the true intergenerational correlation of status is 0.75, as we observe for other societies above, then in Japan the upward bias in the measured intergenerational correlation would be 0.06. Thus even large-scale selective adult adoptions have only a modest and predictable effect on measured persistence rates.

  The original surname lists include many widely held surnames. To narrow the list to surnames that were more closely associated with kazoku and samurai families, only rarer surnames from the list were employed in measuring mobility. Rarer surnames were defined as those now held by fewer than ten people per million (1,270 or fewer people). Unlike China, Japan has considerable surname variety: there are an estimated 110,000 Japanese surnames. So there are reasonable numbers of rare surnames associated with these earlier elites.

  One complication of this approach involves the romanization of Japanese names. Both the sources of the candidate elite surnames give surnames in kanji, the Japanese character system. The source for the modern frequency of these surnames is Public Profiler’s World Family Names, an Internet surname database. Its data for Japan are derived from the surnames associated with forty-five million of the estimated total of fifty-two million households in Japan in 2007. But these names are romanized on the basis of pronunciation, a method that creates difficulties in matching the modern names with the original kanji versions.15 Table 10.5 shows the numerical composition of our two rare surname samples.

  The relative representation of the rare elite surnames was calculated for a variety of high-status occupations in modern Japan: medical researchers, 1989–90;16 attorneys, 1987; corporate managers, 1993; university professors, 2005; and scholarly authors, 1990–2012. In all cases these surnames are overrepresented with respect to their incidence in the general population, as shown in figure 10.2. The average rate of representation is three times the expected rate for the kozaku and 4.3 times the expected rate for the samurai surnames.

  Thus these rare surnames are on average overrepresented in modern Japanese groups of high social status across a broad range of occupations. Interestingly, the samurai surnames, despite their being selected from a genealogy of 1812, are still much more heavily overrepresented in four of the five high-status groups in the modern era than are the kazoku surnames.

  TABLE 10.5. The ra
re surname samples

  FIGURE 10.2. Relative representation of rare surnames among high-status groups, Japan, 1989–2012.

  The expectation is that the rarer these surnames are today, the more likely that the holders are actually samurai or kazoku descendants. Thus the rarer the surnames, the greater their predicted overrepresentation among elites. Figure 10.3 shows relative representation controlling for the frequency of the surnames. Several things stand out. First, the rarer the surname, the higher indeed its relative representation. The rarest surnames are 12–16 times overrepresented, whereas the most common are only 2–4 times overrepresented. Some of this effect may, however, be a statistical artifact. Since these surname frequencies are based on only a sample (though a very large one) of family surnames in Japan, they give an imperfect measure of the stock of each surname in the population at large. Thus the surnames assigned to the rarest groups tend to be those whose true frequency in the population is greater. Experience elsewhere suggests that the true frequency of these rarest surnames is at least a quarter greater than reported.

  FIGURE 10.3. Relative representation of samurai and kazoku surnames by estimated frequency of surname in Japanese population, ca. 2007.

  But the overrepresentation of the rarest surnames is so great that even if their true frequencies were double the reported figure, they would still be greatly overrepresented. This supports the idea that the rarer the kazoku or samurai surnames now, the more likely it is that modern bearers of the name are descendants of kazoku or samurai forebears.