The Son Also Rises Read online

  THE BRAHMIN POPULATION

  The seven Kulin Brahmin surnames have always been well represented among physicians in Bengal. Since Independence, they have accounted, on average, for more than 16 percent of physicians. The Brahmin-surname population share from 1860 to 2011 is estimated as described in the notes to figure 8.3. Figure 8.5 shows the implied relative representation of Kulin Brahmin surnames among physicians in Bengal under British rule and in West Bengal after Independence.

  Relative representation declines from 5.8 times the average in 1860–89 to 4.2 in 1980–2011. This result implies very low social mobility rates. The recent overrepresentation of the Kulin Brahmins among physicians is, however, lower than the overrepresentation of Ashkenazi Jews among U.S. physicians, who have a relative representation of 5.6. So while Kulin Brahmin descendants are an elite, they are less distinctive in Bengal than Ashkenazi Jews are in America. There are plenty of poorer Brahmins.10

  FIGURE 8.4. Relative representation of surnames among physicians and police sergeants.

  FIGURE 8.5. Kulin Brahmin relative representation among Bengali physicians and non-Muslim Bengali physicians, 1860–2011.

  However, as the figure shows, the apparent decline in the relative status of Kulin Brahmins is mostly due to the partition of Bengal at Independence in 1947 and the loss of a large portion of the low-status Muslim population. After Independence, these surnames show little sign of regressing toward average representation among physicians. Only since 2000 has Brahmin overrepresentation declined, and this may just be a blip. During the colonial period, Kulin Brahmin relative representation was rising, though this was mainly because of the relative growth of the poor Muslim population.

  Looking at the representation of Brahmin physicians among only the non-Muslim population (as represented by the dotted line in the figure), the relative representation of Brahmins shows very little sign of regression to the mean in either epoch. Even in the period since Independence, the persistence coefficient is 0.97. Surprisingly, the reservation system in Bengal, which sets aside 28 percent of medical-school places for scheduled castes and tribes, has produced little downward mobility among the Kulin Brahmin surname group since the colonial era.

  As shown below, the reservation system did sharply increase the representation of a group of surnames associated with scheduled castes. What would the rate of downward mobility of the Brahmin surnames have been had the system not been implemented? Assuming that the system caused the Brahmin community to lose access to 28 percent of medical-school places and adjusting the data accordingly, the relative representation in the final period 1980–2011 would rise to 4.1 among the non-Muslim population, which is higher than the rate before Independence.11 The implication is that absent reservation, there would have been no downward mobility among the Brahmin community in Bengal from the mid-nineteenth century to the present. India would be an example of a society with no mobility for some social groups.

  OTHER ELITE HINDU SURNAMES

  Our group of other surnames associated with high status shows a nearly fivefold overrepresentation among physicians in Bengal in the years 1860–89. The implied intergenerational correlation of status for the colonial period differs modestly depending on whether it is calculated for the population as a whole or only for the non-Muslim population. But, as table 8.2 shows, across both the colonial and Independence eras, it is around 0.86, also a high rate (though lower than for the Brahmin surname group). Somewhat lower on the social scale than the Kulin Brahmins, this group seemingly faced more competition for unreserved places at universities. But again, the implied rates of downward social mobility for this group of surnames remain low, even despite the expected effects of the reservation system in reducing their share of physicians. As with the Kulin Brahmin surnames, without the reservation system, the relative representation of these surnames among the non-Muslim population in the period 1980–2011 would be 2.2, very modestly less than the rate of 2.4 in the period 1860–89 under British rule. The underlying rate of social mobility for this group between 1860 and 2011 is consistent with an intergenerational correlation of 0.95 or higher. Thus inherent mobility rates are again very low.

  THE POOREST HINDU GROUPS

  Despite the establishment of the reservation system, surnames associated with the poorest Hindu groups of the colonial era are extremely rare among physicians even now. Among the non-Muslim population, they appear among physicians at 4 percent of the expected rate. They are also greatly underrepresented among lower-status occupations such as police sergeants and subinspectors in Kolkata (see figure 8.4).

  As table 8.2 shows, the implied persistence rate for this group is 1.01 under British rule, implying no upward mobility. Since Independence, the calculated persistence rate has fallen to 0.83–0.85, depending on the reference group. But representation of this surname group among physicians is so low that this change in measured mobility rates may be the result of random chance.

  Despite ample room for improvement of status, these surname groups have benefited little from the reservation system. Some of these surnames, such as Dhanuk, belong to groups which, although poor, did not qualify as scheduled castes because the British did not list them as such in 1931. While at least some Shaw/Shows were among the scheduled castes, many clearly were not. Thus in the list of nearly five hundred recruits to the Kolkata police with the rank of sergeant or subinspector, the four Shaws were all found in the “general,” or unreserved, category. In a sample of medical-school admissions for 2010–11, three of the four Shaws were in the general category.

  THE SCHEDULED CASTE ELITE SURNAME GROUP

  The peculiarity of the scheduled caste surnames identified above is that although all of them figure prominently in the scheduled caste list, they also figure significantly in the list of Bengal physicians from before Independence. Indeed, as figure 8.6 shows, these surnames were already fully represented among physicians relative to their share in the population in the last generation before Independence. Looking just at the share of these surnames among the non-Muslim population, they were at less than half their expected representation in the period 1860–89 but were converging toward proportional representation, with a persistence rate of 0.84.

  The success of this surname group under the reservation system has led to these surnames becoming as overrepresented as many higher-caste Hindu surnames among both physicians and police recruits (see figure 8.2). Because they start just below the mean representation in the first generation after Independence, there is no implied regression to the mean for this group. They are defying the law of social mobility presented in chapter 6, which predicts that all groups regress to the mean.

  This recent overrepresentation of these surnames among physicians, even with respect to the non-Muslim population only, seems to be driven by the reservation system. In a list of recent admissions to medical schools in West Bengal that identifies some students by their reservation-system category, this surname group, accounting for 141 admissions, was at double the average representation for the non-Muslim community. Only 30 percent of this group were admitted to unreserved places; the rest were assigned places reserved for scheduled castes.12 In the absence of reserved places, only fifty-eight surnames from this group would have appeared, and the group would have had a relative representation of only 0.84 instead of 2.04.13

  FIGURE 8.6. The curious history of the scheduled caste surname group.

  These results seem to be driven by the arbitrariness of the original scheduled caste and scheduled tribe lists drawn up by the British, which ended up classifying even moderately prosperous groups as “untouchables” and reserving places for them. These misclassified groups thus gained a disproportionate advantage from the reservation system.14

  THE MIXED HINDU SURNAMES GROUP

  The group of mixed Hindu surnames includes surnames that were elite during the colonial era and showed no tendency then to regress to the mean. Since Independence, these names have tended to diverge from the mean, becomi
ng more elite relative to the general population. But with respect to the non-Muslim population, these surnames show close to average representation among physicians both during the colonial period and since Independence. It is thus not possible to estimate a rate of regression to the mean for them because they already are at the mean.

  This surname group both benefits and suffers from the reservation system. Those not designated as members of scheduled castes have a lower chance of admission to university, but those who are members of scheduled castes have a comparable advantage. Looking at lists of admissions from the two medical schools that made public their admissions in the reservation-system categories, 58 percent of this surname group were admitted to unreserved places. In the absence of reserved places for this group, the relative representation of this group of surnames, compared to other non-Muslim surnames, would have dropped from slightly above 1 to 0.8.

  Social Mobility Rates Without the Reservation System

  The strange pattern of convergence and divergence seen in figure 8.3 and table 8.2 seems to be an artifact of the reservation system for university admissions. Table 8.3 shows the relative representation of each of the six surname groups among physicians first registering in Bengal between 2000 and 2011. Using the cases noted above from universities that reported the reservation-system status of their admitted students, it is possible to estimate the share of reserved-place admissions to medical school for each surname group. Because this sample is small, for one group, the poorest Hindu surnames, there are only four people observed.

  With this information, we can estimate the representation of the various surname groups for the years 2000–2011 had all admission been by open competition. Column 4 of the table shows the implied relative representation in this case. Figure 8.7 shows the estimated relative representation for each group for 1920–2011 without the reservation system.15

  From this counterfactual estimate of relative representation, the implied persistence coefficient between the generation of physicians in the periods 1920–47 and 2000–2011 (seventy years or 2.33 generations later) is estimated. These estimates are shown in the rightmost column of the table.

  These calculations imply that without the reservation system, for Kulin Brahmins, other high-status Hindus, and Muslims, there would have been little or no regression to the mean. The mixed and poorest Hindu surname groups regress toward the mean at a slow rate. For the poorest Hindu surname group, however, the numbers of physicians observed is so low that this result may be spurious.

  The scheduled caste surname group still shows the odd transition from an underrepresented to an overrepresented group among physicians. But the attempt here to control for the effects of the reservation system is only partial: it does not control for the effect of the reservation system on the previous generation, which might have created more middle-class families whose children were better able to compete for unreserved places.

  The effects of the reservation system between 1950 and 1999 cannot be fully inferred. On balance, it may have reduced the persistence rate for the initially high-status groups. But it has also served to increase persistence for a large and growing underclass of Muslims and poor Hindus who are ineligible for scheduled caste status.

  TABLE 8.3. Implied persistence of status with and without the reservation system, medical-school admissions, Bengal, 2000–2011

  FIGURE 8.7. Social mobility by surname type without reservation system, 1920–2011.

  It is also not clear whether the system is doing much to increase these overall slow rates of social mobility. As evidenced by surname distributions, the two-thirds of the population outside the reserved categories in Bengal has seen little change in relative social position over the past two generations. Among the groups included in the reserved categories, a few seem to have reaped disproportional gains, while others seem to have experienced few benefits. Thus despite the intergenerational mobility injected by the reservation system in the short run, the impression from the surname-group analysis is of an overall rate of social mobility close to zero. India seems to be a uniquely immobile society.

  Does Marital Endogamy Explain Low Mobility?

  The social mobility rates for Bengal estimated above are among the lowest observed, lower than those of preindustrial England, Sweden, or China. They are lower than those in highly unequal contemporary societies such as Chile, and are certainly lower than those of the modern United States, England, Sweden, China, Taiwan, Japan, and Korea. What makes India different from these other societies? Does this finding suggest that there is no universal constant of social mobility?

  The hypothesis preferred, in line with the argument of chapter 7, is that Bengal’s low social mobility rates are due to uniquely low rates of intermarriage between members of traditional elites and underclasses in Bengal. Chapter 7 argues that an important source of social mobility is marital sorting driven by the current social and economic status of families. With such sorting, those of high underlying social status tend to marry those of lower underlying status, and their children consequently regress to the mean. If, however, groups of high average status marry only within that group, then, although there is social mobility within the group, the group’s average status will not revert to the mean. So if Indians conform to this pattern of associative marriage and also operate with rigid barriers to marriage across religious and caste lines, then religious and caste groups will not revert to the mean.

  If marriage is endogamous, then there is no mechanism to eliminate underlying differences in the average levels of ability or competence of different castes or religious groups, and they will exhibit little or no social mobility.

  Despite the importance of religion and caste in Indian history and politics, there has been surprisingly little study of intermarriage between different social groups in India in general or in Bengal specifically. As late as the 1960s, caste endogamy still seemed to be the rule for most marriages in Bengal, as seen in a detailed study of a modest-sized town in Bengal in the late 1960s.16 Another study, looking at marriages in rural villages in Karnataka and Uttar Pradesh for the years 1982–95, found that of 905 marriages in the study, none involved couples who differed in their caste status.17 In Hyderabad, among Kayasthas, only 5 percent of marriages were outside the caste even in the period 1951–75.18 Information on the degree of endogamy for marriages in Bengal in the 1970s and 1980s, which produced the most recent crop of physicians, is not readily available.

  TABLE 8.4. Most common female first names among religious groups, Kolkata, 2009

  One source of information on the likely endogamy rate is the 2010 Kolkata voter roll, which gives surnames, first names, and ages of all voters. Many first names are highly specific to the Hindu, Muslim, and Christian/Jewish communities. Table 8.4 shows the most common ten first names for women in each religious group. Women who marry into one of these groups from another group will almost always have different first names from women born in within the group. Also, if families with surnames associated with one group are assimilated into another group, then, as a result of intermarriage and adoption of at least some elements of the culture of the wives, the children again would typically have different first names.

  As table 8.5 shows, the percentage of women in the Kulin Brahmin surname group with non-Hindu first names is extremely small. Because Muslims constitute nearly a quarter of the Kolkata population, this implies that intermarriage rates between Kulin Brahmin men and women of Muslim origin are extremely low, on the order of 0.1 percent. A similar result holds for other high-caste Hindu surnames.

  More women with Muslim surnames have Hindu first names: 0.9 percent. But given the near-total absence of any sign of Muslim women’s marriage into high-caste Hindu groups, if these findings are indicative of marriage alliances, they are likely with lower-caste Hindus.

  Intermarriage between Christians and high-caste Hindus appears to be substantially more common. Christian surnames account for a very small share of the surname stock in Kol
kata, about 0.3 percent, and are mainly Portuguese in origin. Given this small Christian population, the small share of women with high-caste Hindu surnames who have Christian first names is nevertheless suggestive of significant intermarriage.

  An alternative explanation for these female Christian first names may be that high-caste Hindu girls are given Christian first names at birth. The possibility of significant intermarriage between Christians and Hindus is, however, supported by the fact that just over 30 percent of women with Christian surnames have first names that are Hindu. Also, almost 12 percent of women with Christian surnames have a combination of Christian and Hindu first names.

  The first-name and surname evidence suggests almost no intermarriage between the largely poor Muslim community and either Hindus or Christians. Within the Hindu community, first-name evidence does not allow us to determine the degree of marital endogamy within castes because many female first names are common to high- and low-caste groups.

  TABLE 8.5. Female first-name origins by surname group

  One of the few exceptions is the name Munni, found at the rate of 0.007 percent among high-caste surname women and 0.20 percent among other Hindu surname groups. If Munni were distributed representatively in the rest of the Hindu population, maintaining this incidence disparity would require that less than 4 percent of men with elite surnames married women from the general Hindu population. This would indicate a high degree of marital endogamy among elite populations.

  However, the incidence of the name Munni is inversely proportional to the wealth of the group. Among the poorest Hindu surname group, it is found at a rate of 0.9 percent. Thus the absence of the name Munni among women with elite surnames may not reflect general marital endogamy among these caste groups but only a lack of intermarriage between men with high-caste surnames and low-status women. This type of marital sorting would be the same as that observed in societies such as the United States, England, and Sweden and thus would not suffice to explain the lack of social mobility in India.