 Show
Jason Long Department of Business & Economics , Wheaton College , 501 College Ave, Wheaton, IL 60187 , USA Search for other works by this author on: Published: 01 February 2013
Close Navbar Search Filter Microsite Search Term Search AbstractThis paper derives new estimates of social mobility in England and Wales between 1851 and 1901, using a large new dataset of fathers and sons linked across censuses from 1851–1881 and 1881–1901. Mobility rates were substantially greater than has been previously estimated, to the extent that mobility in the 1850s was only slightly less than in the 1970s. The development of mass public education in England after 1870 thus had surprisingly modest effects over the long run. Earnings mobility increased moderately for the first generation under public education (1881–1901), but did not increase over the course of the twentieth century. In assessing the level of equality in a society, it is natural to look first at the distribution of economic resources across the population. A high concentration of income or wealth indicates inequality in economic outcomes. However, it is equally important to consider the rate of social mobility, which indicates the equality not of outcome, but of opportunity. In a society that rewards skill and effort, which vary across the population, outcome inequality will always exist. The impact of that inequality, although, depends critically on the extent of social mobility—in other words, the extent to which occupation, class, and earnings are transmitted across generations and across the work life. Empirical studies of economic equality and distribution have long been a mainstay of the economics literature. Recently, widespread scholarly interest in the empirical study of social mobility has brought our understanding of mobility in contemporary economies in line with our understanding of distribution.1 Economic history, on the other hand, has a long tradition of vigorous study and debate on the long-run trends in distribution, but a relatively less complete empirical understanding of social mobility before the twentieth century.2 In this paper, I use a new dataset of fathers and sons linked from the population censuses of 1851–1881 and 1881–1901 to measure both inter- and intragenerational social mobility in nineteenth-century England and Wales.3 Unlike previous data sources, these data allow for father's and son's occupations to be observed at roughly equal ages. Controlling for life cycle in this way, I find that intergenerational social mobility was markedly greater than previous estimates have indicated. This difference is due to the presence of significant intragenerational mobility, which is measured here for the first time. A comparison with data from the 1972 Oxford Mobility Study (OMS) reveals that although social mobility increased slightly from 1851 to 1972, the increase is much less than has previously been estimated. Indeed, when long run changes in the wage structure are taken into account, it appears that the intergenerational earnings elasticity has changed very little in Britain from the second half of the nineteenth century to the second half of the twentieth. Considering the dramatic institutional change that occurred over this long time period, in particular Britain's large investment in public education, it is striking that the overall rate of mobility in late twentieth-century Britain appears to be largely unchanged from its level in the mid-nineteenth century. 1. Past and present views of Victorian mobilityThe scarcity of historical empirical mobility studies does not reflect a lack of interest in the subject, either currently or among nineteenth-century commentators. This is certainly true of nineteenth-century Britain, where the idea of the “self-made man” and the open society through which his industry, effort, thrift, and ability would propel him were potent conceptions of society at the time. Two famous statements convey a sense of the contemporary debate around these ideas. No doubt the best-known apologist for the rewards of hard work and the possibility of advancement was Samuel Smiles, who wrote in 1859 in his most famous work, Self-Help, that “great men of science, literature, and art—apostles of great thoughts and lords of the great heart—have belonged to no exclusive class nor rank in life. They have come alike from colleges, workshops, and farmhouses—from the huts of poor men and the mansions of the rich” (Smiles 2002, p. 7). Perhaps the most famous contradictory position was voiced by John Stuart Mill, who wrote in Principles of Political Economy of a “demarcation between the different grades of laborers, as to be almost equivalent to an hereditary distinction of caste” (Mill 1904, p. 238). Subsequent scholars have emphasized the importance of openness and mobility in Britain's growth experience. David Landes (2003, pp. 48–50, 128–130, 544–547) cites limited class rigidities in eighteenth-century Britain relative to continental Europe as a positive factor in Britain's early industrialization. Mancur Olson (1982, pp. 82–87), on the other hand, blames Britain's sluggish growth after World War II in part on its strong class rigidities compared with the more war-ravaged and hence more fluid continental economies. Until recently, there has been much more conjecture and speculation regarding social mobility in nineteenth-century Britain than broad empirical analysis. In the USA, Thernstrom's (1964, 1973) well-known studies of social mobility in Newburyport and Boston helped to foster a relatively active agenda of historical mobility studies. In Britain, on the other hand, historians have produced a less systematic body of work, tending to focus on recruitment to more specific occupational groups and on more narrowly local studies.4 Several studies from the 1990s took important steps toward providing a more comprehensive empirical picture of mobility in nineteenth-century England. Andrew Miles (1993, 1999) and David Mitch (1993) have provided most of what we know about social mobility patterns more broadly for nineteenth-century England through their meticulous studies of marriage registries.5 Miles examines 10,000 English marriage registries from 1839 to 1914 and shows that 38 percent of grooms over the period were in a different occupational class than their father at the time of marriage. He concludes that “in terms of its inhabitants' relative life chances [Victorian and Edwardian England was] a profoundly unequal society” (Miles 1999, p. 177). Mitch compares father/son mobility with father-in-law/son-in-law mobility (endogamy) and finds that the latter process was somewhat more fluid than the former. Nearly all studies of social mobility in Europe in the nineteenth century rely on data from marriage registries. The English data are typical. Beginning with the 1836 Registration of Births, Deaths, and Marriages Act, Anglican churches in England were required to record the occupations of both partners and their parents at the time of marriage. These records provide a snapshot of father and son's occupations that can be used to measure the rate of intergenerational occupational mobility.6 The marriage registers are subject to well-known limitations, which both authors readily acknowledge. Of course, they exclude the non-marrying population. In addition, they only include information from Anglican ceremonies. Neither of these biases is likely to be particularly serious early in the nineteenth century; only about 10 percent of 45-year-old males did not marry, and most weddings were Anglican. By 1914, however, over 40 percent of marriages occurred outside the Anglican Church, so late in the period the potential bias is more serious. Furthermore, the marriage registry sample includes information from only ten registration districts. The districts were selected to maximize the variation in economic and social characteristics; still, there were never fewer than 600 districts in England and Wales during the time period (there were 624 registration districts at the time of the 1851 census), so including data from only 10 districts should be expected to limit significantly the representativeness of the sample for England and Wales as a whole. Each of the data limitations described above calls into question the degree to which the marriage registry sample is well representative of the population of England and Wales. But certainly, the most serious limitation of the source derives directly from the nature of each individual record. In observing father and son's occupation at one point in time—the time of the son's marriage, when the son is typically a relatively young adult—it is doubtful that one observes true occupational mobility over the course of two generations. It is quite likely that many sons would have changed job and social class after marriage and that upward mobility would dominate downward as sons gained skills and experience. This “snapshot problem” means that mobility measures derived from marriage registries would be likely to underestimate the true degree of intergenerational mobility, with estimates of upward mobility being particularly downward biased.7 This difficulty is directly related to the other fundamental and serious limitation of the marriage registries as a source for studying mobility: they offer no information at all on intragenerational mobility, that is, mobility over the course of the individual's work life. In this, the English marriage registry data are like virtually all data sources used to study social mobility in the nineteenth century. Without including information on specific people over time, nothing can be learned about the prevalence of individual, career mobility.8 The implication is that virtually everything we know about nineteenth-century mobility, at least generally and quantitatively, relates to changes across generations. Up to now, we have no rigorous quantitative information about mobility over the individual career for England and Wales as a whole.9 2. Analyzing mobility with linked census dataDespite their limitations, marriage registers have had the one clear advantage of being the only source available in large numbers for the entire country that contain occupational information for fathers and sons. Until now, they have been the only means available for the broad empirical analysis of social mobility, and then only intergenerational mobility. Recently, however, a variety of extremely fruitful genealogical databases and research tools have been made publicly available that allow the large-scale linkage of individuals across censuses in Britain between 1841 and 1901. With these resources, I have constructed a large, nationally representative dataset of 54,218 males covering three generations of household heads from 1851 to 1901. This data source allows each of the main shortcomings of the marriage registry data to be addressed. Most importantly, by tracking individuals over a long time period, intragenerational mobility can be observed, and the life-cycle problem in measuring intergenerational mobility can be overcome by comparing fathers and sons at comparable ages. In addition, the dataset is large and covers essentially the entirety of England and Wales. Three sources are used to perform the linkage: (i) a computerized 2 percent sample of the 1851 census, (ii) a computerized version of the complete-count 1881 census, and (iii) the complete-count 1901 census, accessible through Ancestry.com, a web-based genealogical research service.10 The first stage of the data creation process was the nominal linkage of 28,474 males from the 1851 census sample to the 1881 complete count census. Complete details of the linkage procedure and data construction process, including the matching algorithm, potential sources of bias, and expected versus actual linkage success rate, have been reported elsewhere and will not be repeated here.11 Briefly, individuals were linked based on first and last name and county, parish, and year of birth—information that should, barring error, remain constant across censuses. Some leeway in the matching algorithm was allowed for small discrepancies in reporting personal information across censuses 30 years apart. Names were allowed to vary slightly, as long as they matched phonetically, and reported age in 1881 was allowed to deviate by up to 5 years from the expected value based on reported age in 1851.12 This linkage procedure produced a sample that is well representative of the young male population of England and Wales. The sample is younger than the population as a whole for the simple reason that men had to have survived 30 years to be found in the 1881 census. Table 1 illustrates the representativeness of the data by comparing the linked individuals from the 2 percent sample of the 1851 census with the entire sample. Observable characteristics are compared across the two groups used to measure social mobility: sons under the age of 20 and males between the ages of 20 and 35. Table 1. Representativeness of linked census data
Notes: All figures are percentages except where noted. Columns 1 and 2 compare all sons younger than age 19 in the 1851 2 percent census sample with those sons successfully linked into the 1881 census. Columns 3 and 4 compare all males between the ages of 20 and 35 in the 2 percent sample with those males successfully linked into the 1881 census. For sons, Occupation, 1851 refers to father's occupation; for males of age 20–35, it is their own occupation. Sources: 1851 census 2 percent sample and new sample of linked individuals. Table 1. Representativeness of linked census data
Notes: All figures are percentages except where noted. Columns 1 and 2 compare all sons younger than age 19 in the 1851 2 percent census sample with those sons successfully linked into the 1881 census. Columns 3 and 4 compare all males between the ages of 20 and 35 in the 2 percent sample with those males successfully linked into the 1881 census. For sons, Occupation, 1851 refers to father's occupation; for males of age 20–35, it is their own occupation. Sources: 1851 census 2 percent sample and new sample of linked individuals. Because the census records households together and because the 1851 sample preserves this household structure, it is simple to connect the young linked males who were sons living with their family in 1851 with their fathers. This provides the basic structure necessary to observe intergenerational mobility from 1851 to 1881, and it does so in such a way that both father and son are observed as mature adults, with approximately equal ages, at approximately the same point in the life cycle. For the 12,647 father/son pairs in which the son is aged 0–19 years and living with his family in 1851, the average age of the father in 1851 is 41.5 years, and the average age of the son 30 years later in 1881 is 38.0 years. An average age difference of only 3.5 years should have a negligible impact on observed mobility considering the advanced age of both father and son at the time of observation. It is impossible to compare directly this age pattern with that of the marriage registry data, as only two-thirds of the registries record the ages of groom and father; in fact, for the earliest time period (1839–1854), only one in three grooms recorded their age. We do know, however, that these sons typically would have been fairly young at the time of their first marriage. Wrigley and Schofield (1989, p. 255) give the average age at first marriage for males in England from 1840–1849 as 25.3 years. Given both the large age gap that would have existed between father and son, and the early career stage of the son at time of marriage, the age pattern of the linked census data offers a substantial improvement in controlling for life-cycle effects relative to the marriage registry data. Again, this is true for the measurement of intergenerational mobility from father's mature occupation to son's mature occupation; in other words, the total or overall amount of mobility from one generation to the next. Insofar as our interest in mobility is primarily motivated by a desire to understand and quantify the degree of equity or fairness in the economy, this is probably the single most important mobility metric. But, it is not the only one. It is also informative to measure the extent to which father's occupation influences the son's first occupation, or “entry point” into the labor market, and the extent to which individuals experience mobility from that entry point to their subsequent mature occupation. Neither the linked census data nor the marriage registry data can be used to measure all three types of mobility. The linked census data can reveal intragenerational mobility and overall intergenerational mobility, but not mobility to first job. This suggests that perhaps the best use of the marriage registry data is not to measure overall intergenerational mobility at all, but rather to measure mobility to first job. In this sense, results from the marriage registry data may be seen to complement the results presented here from the linked census data. To measure mobility over the individual work life, as opposed to across generations, older individuals with meaningful occupations in 1851 are used. This group comprises 7,790 males between the ages of 20 and 35 in 1851. Intragenerational mobility is revealed by observing their occupation in 1851 and again 30 years later in 1881. The final stage of the linkage process adds a third generation to the data. Twenty thousand two hundred and sixty nine sons were included from the households of the linked males in 1881, and 8,677 were linked into the 1901 complete count census using Ancestry.com's web-based genealogical research service. These individuals are used to measure intergenerational mobility from 1881 to 1901. Table A1 summarizes the three datasets used to construct these mobility measures, and Table A2 uses the households of three successive generations of males from 1851, 1881, and 1901 to illustrate the nature of the linked census data. The primary measure of intergenerational mobility used here is mobility of occupational class, commonly referred to as social mobility in the literature. This metric is most directly available from the censuses, which contain information on occupation, but not earnings. Although earnings is the measure most commonly used in the economics literature on mobility, there are advantages to using occupation (as is the norm in the sociology literature). One of the principal empirical difficulties in the study of earnings mobility is obtaining a true measure of permanent income in the face of frequent transitory income shocks. As shocks often occur without job changes, occupation should be less affected by such disturbances. Further, compared with a simple earnings measure, occupation and social class capture more dimensions of an individual's experience that may be related to interpretations of social mobility such as prestige in the community, autonomy in the workplace, manual versus non-manual labor, place of work, and so on. Finally, and most importantly for the current study, in order to compare results with those of Miles, Mitch, and Goldthorpe, it is necessary to measure mobility in terms of occupational class, as each of those authors does. A clear disadvantage of using occupation is that, unlike a simple, continuous metric like earnings, occupation somehow must be coded and classified for it to be useful. Here, I follow the standard practice of applying W. A. Armstrong's classification system, which is based on the Registrar General's 1921 and 1951 classification schemes (Armstrong 1972). Every individual is assigned to one of five ranked social classes according to his occupation as recorded in the census enumerator's book: I—Professional, II—Intermediate, III—Skilled, IV—Semiskilled, and V—Unskilled.13 This is a classification system based solely on occupation, and although there are additional components of social class, occupation is nearly always considered to be of central importance in determining an individual's class. Armstrong (1972, p. 202) deems occupation “our best guide to social class” and cites Thernstrom (1964, p. 84): “Occupation may be only one variable in a comprehensive theory of class, but it is the variable which includes more, which sets more limits on the other variables, than any other criterion of status.” Armstrong's aim for the classification system is to “ensure that each category is homogeneous in relation to the basic criterion of the general standing within the community of the occupations concerned.” Under Armstrong's system, each occupation is coded according to the Registrar General's classification scheme, with several modifications made to minimize anachronism. The most important modification is that, regardless of job title, all employers of twenty-five or more are included in Class I, and all individuals with Class III or IV occupations employing at least one person other than a family member are included in Class II. Empirically, this scheme correlates well with other indicators of social class. In this sense, what is referred to here as “social class” is essentially synonymous with what is often referred to in the social sciences as socioeconomic status. According to occupational/industrial pay estimates compiled by Jeffrey Williamson (1980, 1982), the average wage premium in 1851 for each class relative to the next lowest class was 7 percent for Class IV, 33 percent for Class III, 81 percent for Class II, and 45 percent for Class I.14 Furthermore, Armstrong (1972, p. 212) demonstrates that job class, defined according to this system, is positively correlated with the employment of servants and negatively correlated with the incidence of shared accommodation. Apart from a consideration of its relative advantages and disadvantages, this classification system is necessary to achieve the desired comparability across studies. Both Miles and Mitch derive social mobility in terms of the Armstrong/Registrar General occupational classes, and the twentieth-century data used to assess the long-run trend can be classified readily with the 1951 Registrar General's system. 3. Mobility within and across generationsTables 2–4 show the inter- and intragenerational mobility rates by class of origin for the period 1851–1901. The tables show column percentages, typically interpreted as mobility rates. Table 2 shows the percentage of sons attaining each occupational class in 1881, by class of origin (father's class) in 1851. Table 3 shows the percentage of sons attaining each occupational class in 1901, by class of origin in 1881. Table 4 shows the percentage of males attaining each occupational class in 1881, by class of origin (own class) in 1851. The elements off the main diagonal (upper left to lower right) represent total mobility—all individuals whose “destination” class is different from their “origin” class. Because the occupational class scheme is strictly ordered from I (highest class) to V (lowest class), it is straightforward to identify elements in the table above the main diagonal as representing upward mobility and those below it as representing downward mobility. Given these definitions, we see that across generations from 1851 to 1881, the rate of total mobility is 50.1 percent, the rate of upward mobility is 26.8 percent, and the rate of downward mobility is 23.3 percent. From 1881 to 1901, those same figures are 48.3 percent, 26.8 percent, and 21.5 percent, respectively. The rate of total intragenerational mobility, 43.6 percent, is slightly lower than the rate of total intergenerational mobility, but at first glance, it would still appear to be a reasonably high rate of mobility. It is certainly high enough to support the notion that measuring sons' mobility at the time of their marriage is likely to underestimate significantly the true rate of total intergenerational mobility. This is particularly true with respect to upward mobility, as the rate of upward intragenerational mobility, 25.4 percent, is markedly greater than the rate of downward mobility, 18.2 percent. The post-marriage mobility hypothesis can be tested directly by repeating the intragenerational mobility analysis on the set of married men younger than 35 found in the linked census data. These results are reported in Table A3. The rates of total, upward, and downward mobility—40.5, 22.6, and 17.9 percent, respectively—are similar to the overall intragenerational mobility rates. Table 2. Intergenerational mobility, 1851–1881: column percentages (mobility rates)
Note: Sons aged 0–19 in 1851. Table 2. Intergenerational mobility, 1851–1881: column percentages (mobility rates)
Note: Sons aged 0–19 in 1851. Table 3. Intergenerational mobility, 1881–1901: column percentages (mobility rates)
Note: Sons aged 10–19 in 1881. Table 3. Intergenerational mobility, 1881–1901: column percentages (mobility rates)
Note: Sons aged 10–19 in 1881. Table 4. Intragenerational mobility, 1851–1881: column percentages (mobility rates)
Note: Males aged 20–35 in 1851. Table 4. Intragenerational mobility, 1851–1881: column percentages (mobility rates)
Note: Males aged 20–35 in 1851. Several specific features of the mobility tables are worth emphasizing. First, mobility appears to be lowest out of skilled (class III) occupations. This apparent immobility is largely due to the excessive size of this category, which is one of the principal drawbacks of the Armstrong/Registrar General classification scheme. Indeed, Table 2 shows that over half of all sons held a class III occupation in 1881, and 47 percent of fathers held a class III occupation in 1851. Using the Armstrong scheme alone masks much of the meaningful mobility that occurred across generations during these 30 years. Occupational mobility within class III was substantial: fully 71 percent of “immobile” class III sons had a different occupation in 1881 than their father in 1851. Williamson's occupational/industrial pay estimates can be used to impute wages for these individuals by occupation, revealing additional information on mobility within the otherwise undifferentiated class III.15 Of the sons with a different class III occupation than their father, 61 percent had an occupation that paid on average at least 10 percent more, in real terms, than the average wage associated with their father's occupation. Using the wage information to divide the class III occupations into five subgroups, we see a great amount of earnings mobility occurring within the class. Table A4 shows the full results. Total earnings mobility within class III is 41 percent, of which 24 percent is upward and 17 percent is downward. Combining these expanded subclasses of the skilled occupational group with the existing information on classes I, II, IV, and V into a nine-class scheme, although somewhat ad hoc, does provide perhaps the fullest possible measure of true intergenerational socioeconomic mobility over the time period. The resulting picture is one of substantially greater mobility than has been observed previously. According to this scheme, fully two-thirds of sons occupied a different socioeconomic strata in 1881 than their fathers in 1851,36 percent at a higher level and 30 percent at a lower level. Full results are shown in Table 5. Table 5. Intergenerational mobility, 1851–1881: expanded class categories
Notes: Sons aged 0–19 in 1851. Class III: skilled occupations expanded using wage information from Williamson (1980, 1982). All wages were calculated at 1851 prices. Table 5. Intergenerational mobility, 1851–1881: expanded class categories
Notes: Sons aged 0–19 in 1851. Class III: skilled occupations expanded using wage information from Williamson (1980, 1982). All wages were calculated at 1851 prices. The second feature of the mobility table worth noting is that upward mobility is substantial, but not across the entire distribution of classes. Mobility from classes IV and V into class III is relatively high. For example, only a third of young men in unskilled jobs remain there, while 40 percent move up into a skilled occupation. However, very few men moved into class I or II white-collar jobs from the ranks of the manual workers, either across or within generations. In this sense, pessimists with respect to mobility in Victorian Britain, although perhaps understating the overall level of mobility, appear to have accurately described the difficulty of moving into the white-collar ranks from manual backgrounds. Given the substantial rate of intragenerational mobility shown in Table 4, there is every reason to believe that previous estimates of mobility based on the marriage registries are too low. The results reported in Tables 6 and 7 bear this out. These tables show the direct comparison of intergenerational mobility rates derived from the 1851–1881 linked census with the rates derived by Miles from the sample of marriage registries from 1859–1874.16 Total mobility is revealed to be 36 percent greater in the linked data than in the marriage registries: 47.5 versus 34.8 percent. As we would expect given the relatively high rate of upward intragenerational mobility revealed in the linked census data, most of this discrepancy is driven by differing rates of upward mobility; the linked census data show 57 percent more upward mobility than do the marriage registries (27.8 versus 17.7 percent). The difference between new and old estimates of mobility is even greater, if we compare the five-class marriage registry data to the nine-class linked census data. The latter reveals twice as much upward mobility (36 versus 18 percent) and almost twice as much total mobility (66 versus 35 percent) as the former. Table 6. Intergenerational mobility in Victorian Britain: linked census data vs. marriage registries
Notes: Column percentages (mobility rates) shown in both tables. Table 6. Intergenerational mobility in Victorian Britain: linked census data vs. marriage registries
Notes: Column percentages (mobility rates) shown in both tables. Table 7. Summary of mobility comparisons
Notes: M is total mobility, U is upward mobility, and D is downward mobility. G2 is the likelihood ratio χ2 statistic. All G2 statistics are significant at the 1 percent level. For (3), males from the OMS ages 30–49 were used; for (4), males aged 30–39 were used. Table 7. Summary of mobility comparisons
Notes: M is total mobility, U is upward mobility, and D is downward mobility. G2 is the likelihood ratio χ2 statistic. All G2 statistics are significant at the 1 percent level. For (3), males from the OMS ages 30–49 were used; for (4), males aged 30–39 were used. To compare mobility systematically between two tables, it is helpful to consider a single metric that summarizes the difference in mobility between the two tables and that is testable for statistical significance. For examining what sociologists call social fluidity or exchange mobility in a labor market—the rate of mobility controlling for differing occupational distributions between two times or places—the metric should also be invariant to changes in the marginal frequencies of the various classes. Sociologists frequently refer to a distinction between “absolute mobility,” that is de facto, observable mobility, and “relative mobility,” or the mobility of one social group compared either with some predefined norm or with the mobility of another group or class. One statistic that summarizes mobility for a table and quantifies relative mobility is provided by Altham (1970) for two tables P and Q, each with r rows and s columns: where pij and qij are the cell counts for row i and column j of tables P and Q, respectively. The intuition behind the statistic is straightforward. In a simple 2 × 2 table, the degree of mobility can be summarized by the cross-product ratio: p11p22/p12p21. As this measure is invariant to scalar multiplications of rows and columns, it is not affected by differences in marginal frequencies between two tables (Altham and Ferrie 2005), which in this case implies that the statistic is not affected by different occupational structures across two economies. In a table with more than two rows and columns, there will be several cross-product ratios; the “Altham statistic” simply aggregates the impact of each of the cross-product ratios for the two tables. The metric d(P, Q) tells us the distance between tables P and Q. Tables indicating a very similar pattern of mobility in two labor markets will be “close” to each other, with a small value of d(P, Q) and tables indicating very different patterns of mobility will be “far” from each other, with a large value of d(P, Q). A simple likelihood ratio χ2 statistic G2 (Agresti 2002, p. 140) with (r–1)(s–1) degrees of freedom can then be used to test whether the matrix Θ with elements θij = log(pij/qij) is independent; if we can reject the null hypothesis that Θ is independent, we essentially accept the hypothesis that d(P, Q) ≠ 0, so the degree of association between rows and columns differs between table P and table Q. The statistic tells us only how different mobility is across tables; it does not tell us which table has more mobility. That can be determined by calculating d(P, J) and d(Q, J), where J is a matrix of ones, representing total independence of rows and columns. If d(P, Q) > 0 and d(P, J) > d(Q, J), we can safely conclude that mobility is greater in table Q—in other words, mobility is closer in Q than in P to what we would observe under independence of rows and columns, in which the occupation of a father provides no information in predicting the occupation of his son.17 The first two rows of Table 7 present the results from the Altham test applied to the two mobility tables in Table 6. The initial result of greater mobility in the linked census data (table P) than in the marriage registry data (table Q) is confirmed: Q is farther from independence than P (the distance statistic is 62.6 for Q versus 44.1 for P), and the distance between P and Q is large and statistically significant. 4. British mobility over the long runThe analysis in Section 3 makes clear that social mobility in Victorian Britain was greater than has previously been thought. Mobility, particularly upward mobility, over the work life was common, so earlier estimates that failed to control for life-cycle differences between father and son underestimated mobility. Just how mobile a society Victorian Britain was, however, is not clear in isolation. Was recently industrialized Britain the fluid, open society extolled by Smiles or the immobile, closed society of Mill? Comparing mobility in Britain in the nineteenth century with mobility in Britain in the twentieth century reveals how, if at all, mobility has changed over the long run and contextualizes these new results for the nineteenth century. The standard source for empirical social mobility research in postwar Britain is the 1972 Oxford Mobility Study (OMS). Ten thousand three hundred and nine randomly sampled men were interviewed and asked questions about their occupations at various stages of their working lives; their education, training, and formal qualifications; the occupations, education, and training of various family members (including their fathers); and supporting details such as place of residence, income, political affiliations, and so on.18 Of the 10,000 records in the data, only information on males between the ages of 30 and 49 at the time of the survey is used, to ensure an age structure that is comparable to that of the linked census data. The respondent's occupation at the time of the survey is taken as the son's occupation, and the occupation that the respondent reported his father to have had when the respondent was 14 years old is taken as the father's. For the OMS, occupations were classified according to a seven-class scheme by aggregating categories from the Hope–Goldthorpe occupational scale.19 To make the results from the OMS data comparable to the nineteenth-century data, I have reclassified fathers' and sons' occupations according to the 1951 Registrar General's five-class system, identical to the system used by Armstrong and employed in Section 3. Although comparing occupations and classes across a century inevitably will be imperfect, the nature of the data and the sources used to classify the occupations are in fact reasonably well suited to the task. The OMS occupational information is carefully recorded in such a way that translation into the Registrar General's classification system is straightforward.20 This classification system is separated from the OMS data by only two decades, and, for all its shortcomings, it remains the standard source for classifying nineteenth-century occupations as well. So, in fact, the source that must be used to categorize the nineteenth-century data is in fact probably more naturally suited to classifying the twentieth-century OMS data. The upshot is that comparing occupations and class between the linked census data and the OMS by way of the Registrar General's classification system should be seen as no more problematic than using that system to categorize the occupational information in the nineteenth-century data in the first place. Results from the comparison are presented in Tables 7 and 8. Total mobility is nine percentage points higher in 1972 than 100 years earlier, with virtually the entire difference due to a higher rate of upward mobility: 35.5 versus 26.8 percent. Upward mobility out of the manual classes (III–V) was substantially greater in 1972 than in 1881, with a particularly dramatic difference at the very bottom of the scale. When comparing mobility tables across a 100 year time span that encompasses a second industrial revolution, two world wars, and the rise of the service economy, it is particularly important to control for changes in the occupational distribution to gage the fluidity of the labor market; that is, the relative life chances of individuals from different class backgrounds. The Altham test provides this control. Table 7 shows that the change in mobility from 1881 to 1972 was not driven by changing occupational structure—underlying fluidity actually increased. The 1881 table is farther from independence than the 1972 table, and the distance between the two tables is statistically significant. Table 8. Mobility in Britain, 1851 to 1972: linked census data vs. Oxford Mobility Study
Notes: Column percentages (mobility rates) shown in both tables. Table 8. Mobility in Britain, 1851 to 1972: linked census data vs. Oxford Mobility Study
Notes: Column percentages (mobility rates) shown in both tables. Clearly, the data indicate that intergenerational mobility across occupational classes increased from 1881 to 1972.21 However, given the dramatic change in the structure of wages that occurred over this long period of economic change, it is reasonable to wonder whether and how the economic significance of class mobility changed as well. A simple comparison of income and occupational class from the OMS data with Williamson's nineteenth-century wage estimates indicates the extent of the differences across time. As has already been mentioned, the average wage premium in 1851 for each class relative to the next lowest class was 7 percent for Class IV, 33 percent for Class III, 81 percent for Class II, and 45 percent for Class I. In the OMS data, those premia are much smaller overall: only 17, 13, 40, and 19 percent, respectively.22 The fact that, excepting Class V, occupational classes in the twentieth century were much closer together in terms of earnings than were classes in the nineteenth century implies that the economic meaning of class mobility might be quite different across the two time periods. In particular, measuring mobility by class transitions would understate earnings mobility for the nineteenth century relative to the twentieth. To address this concern, it is necessary to construct an estimate of intergenerational earnings mobility for the nineteenth-century data. As the censuses do not reveal the earnings of individuals, no direct measure can be estimated. The best that can be done is to impute earnings to individuals based on their occupation or occupational class. Williamson's earnings estimates are used here for this purpose. There are well-known problems with these estimates, primarily having to do with a handful of professional occupations, especially solicitors and barristers, surgeons and doctors, and engineers. The wage information for these occupations is derived from a small number of sources and demonstrates extreme variation across time periods (Jackson 1987; Feinstein 1988). For the limited purposes of this study, these problems can be dealt with simply enough by omitting these occupations in the construction of the average wage for each class and year. This method preserves some of the important advantages of the Williamson wage data relative to other sources, particularly their consistent construction across the decades from 1851 to 1901.23 The occupations used for each class, along with the average wage for each relevant year and the class averages from the OMS data, are shown in Table 9. Wages are imputed to individuals by class rather than by occupation to facilitate comparison between the nineteenth-century data—for which there are only 13 occupations or occupational groups to use—and the OMS, which contains information on 222 unique occupations. Table 9. Average annual wages by class
Table 9. Average annual wages by class
With these imputed wages in hand, it is straightforward to estimate what is referred to in the literature as the intergenerational elasticity (IGE) of earnings. The standard statistical model is a simple linear relationship of father's and son's earnings: (1) where t is an index of generations and y is a measure of permanent income. β is interpreted as the IGE of earnings. The specification is purposefully sparse in order that β captures the full association of father's and son's earnings, regardless of the channels through which the association operates. Controls for father's and son's age may also be included. Table 10 shows the results of estimating Equation (1) on all three datasets: 1851–1881, 1881–1901, and the 1972 OMS data. In this case, earnings for each individual is the average wage for his occupational class. Because the OMS does not report father's income, income is imputed to each father according to his occupational class, using the average class wages calculated from the sons' 1972 income information. To facilitate comparison with the OMS, a similar procedure is used to impute income in the two linked census datasets: models (2) and (4), rather than using all the available class/wage information from the Williamson data, use only the 1881 class/wage averages across both years. Models (1) and (3) impute wage to fathers and sons using the class/wage average from their own census year.
Notes: All specifications include age controls. Models (1) and (3) use wages imputed from father's and son's own census year. Models (2) and (4) impute wages to both father and son with 1881 class average wages. Model (5) imputes wages to fathers and sons with 1972 class average wages.
Notes: All specifications include age controls. Models (1) and (3) use wages imputed from father's and son's own census year. Models (2) and (4) impute wages to both father and son with 1881 class average wages. Model (5) imputes wages to fathers and sons with 1972 class average wages. The striking result from Table 10 is how similar the estimate of intergenerational earnings elasticity is across the long time period considered. Although earnings mobility did increase from 1851 to 1972, the overall increase was quite small, and it appears to have come entirely during the second half of the nineteenth century, rather than over the course of the twentieth.24 Indeed, these nineteenth-century IGE estimates are very much in line with the findings of the numerous studies that explore intergenerational earnings mobility in modern Britain. Corak (2006) surveys this literature and reports a range of estimates of β of 0.22–0.42 for OLS estimates and 0.50–0.58 for IV estimates.25 5. ConclusionsSocial mobility in Britain from 1851 to the turn of the century was greater than previous research has indicated. High rates of work-life mobility that have previously been unobservable result in a significant downward bias to estimates of intergenerational mobility that fail to account for differences in the life cycle of fathers and sons observed in a point-in-time snapshot. Insofar as previous low estimates of social mobility have been used to characterize Victorian Britain as an unequal, immobile society, our perception changes in favor of a more charitable view of the society as one of substantial opportunity across the socioeconomic spectrum. Although it still appears that the frequency of dramatic upward moves from the lower, manual ranks into the professional and white-collar classes was low, overall a picture of higher mobility emerges. Interestingly, and perhaps surprisingly, by one measure at least, intergenerational mobility of earnings barely changed at all over the decades from the mid-nineteenth century to the late twentieth. While it seems clear that intergenerational social mobility across occupational classes increased from the time of the 1851–1881 censuses to the 1972 Oxford Mobility Study, it appears that earnings mobility has been roughly constant. I estimate IGE for 1851–1901 to be between 0.26 and 0.37 depending on time period and methodology. This range is extremely close to my estimate for the OMS data, 0.32, and to OLS estimates for late twentieth-century Britain, which typically range from 0.2 to 0.4. It is not entirely clear what should be expected a priori regarding the trend in mobility in Britain over this long time period. Miles (1999) finds that social mobility trended moderately, but consistently upward from 1839 to 1914. On the other hand, Erikson and Goldthorpe (1992) find a largely constant rate of social fluidity in Britain and other industrialized countries between the 1940s and the 1970s. These studies deal exclusively in class mobility, and their results are consistent with the finding here of increased class mobility over the long run. Nonetheless, given the dramatic changes that occurred in the British economy and in British institutions, we might expect to see a more pronounced increase in earnings mobility over this long time period. In particular, from the mid-nineteenth century to the late twentieth, Britain went from being a noteworthy laggard in terms of public education to adopting the norms of the developed world in providing mandatory, universal state-funded education through high school and subsidized post-secondary education thereafter. Standard models of human capital investment and accumulation over two generations provide clear predictions regarding how structural economic differences should influence intergenerational mobility.26 Among the characteristics that are expected to result in a more mobile economy are (i) lower heritability of “intrinsic” human capital, (ii) less productive avenues for human capital investment, (iii) lower earnings return to human capital in general, and (iv) more progressive public investment in children's human capital. Although careful analysis of factors (i)–(iii) for Britain from the 1850s to the 1970s is beyond the scope of the current study, factor (iv) does lend itself at least to preliminary consideration. Beginning with a series of Education Acts in 1870, Britain dramatically increased its public investment in education, causing a gradual increase first in primary school attendance and later in secondary education. Table 11 shows enrollment rates for Britain in comparative perspective from 1830 to 1910. Britain was significantly behind the USA, France, and most other industrial economies in public investment in education in the nineteenth century. As it closed this gap, rising rates of social mobility would be expected. Indeed, some evidence can be seen here that the Education Act of 1870 did have an effect on earnings mobility, which increased from the 1851–1881 cohort to the 1881–1901 cohort. However, given the similarity of IGE estimates for 1881–1901 and modern data, it appears that Britain experienced no overall increase in earnings mobility over the course of the twentieth century. In light of this finding, it may be concluded that the development of mass public schooling in Britain had a surprisingly modest effect over the long run, at least in terms of intergenerational economic mobility. Table 11. Enrollment rates in primary schools: students enrolled per 1,000 children of ages 5–14
Table 11. Enrollment rates in primary schools: students enrolled per 1,000 children of ages 5–14
FundingThis work was supported by the National Science Foundation under Grant No. 0517925. AcknowledgementsI would like to thank seminar participants at Harvard, Oxford, the London School of Economics, and the Washington Area Economic History group; participants at the Exeter Economic History Society conference and the Austin Economic History Association Meeting; and Joseph Ferrie, David Mitch, Joyce Burnette, and the editors of this journal for helpful comments. I gratefully acknowledge the hospitality of the Oxford Economics Department and Nuffield College. Valuable research assistance was provided by Jamyang Tashi, Anthony Abakisi, and Matthew Busch. References
, . Intergenerational economic mobility in the United States, 1940 to 2000 , Journal of Human Resources , 2008 , vol. 43 (pg. 139 - 72 ) . , Categorical Data Analysis , 2002 Hoboken, NJ Wiley . The measurement of association of rows and columns for an r x s contingency table , Journal of the Royal Statistical Society , 1970 , vol. 32 (pg. 63 - 73 ) , . Contingency table comparisons , 2005 Unpublished . . The use of information about occupation , Nineteenth Century Society , 1972 Cambridge Cambridge University Press (pg. 191 - 310 ) , , . , Parents and Children: Incomes in Two Generations , 1983 London Heinemann Educational Books , . An equilibrium theory of the distribution of income and intergenerational mobility , Journal of Political Economy , 1979 , vol. 87 (pg. 1153 - 89 ) , . Human capital and the rise and fall of families , Journal of Labor Economics , 1986 , vol. 4 (pg. S1 - 39 ) , . , . Recent developments in intergenerational mobility , Handbook of Labor Economics , 2011 New York Elsevier (pg. 1487 - 541 ) Volume 4, Part B , , , . . Changes in intergenerational mobility in Britain , Generational Income Mobility in North America and Europe , 2004 Cambridge Cambridge University Press (pg. 122 - 46 ) . Do poor children become poor adults? Lessons from a cross country comparison of generational earnings mobility , Research on Economic Inequality , 2006 , vol. 13 (pg. 143 - 88 ) , , . Intergenerational mobility in Britain , The Economic Journal , 1997 , vol. 107 (pg. 47 - 66 ) , . , The Constant Flux: A Study of Class Mobility in Industrial Societies , 1992 Oxford Clarendon Press . The rise and fall of the Williamson curve , Journal of Economic History , 1988 , vol. 48 (pg. 699 - 729 ) . , Social Mobility and Class Structure in Modern Britain , 1987 Oxford Clarendon Press , . , The Social Grading of Occupations: A New Approach and Scale , 1974 Oxford Clarendon Press . . Intergenerational mobility for whom? The experience of high and low earnings sons in international perspective , Generational Income Mobility in North America and Europe , 2004 Cambridge Cambridge University Press . The structure of pay in nineteenth-century Britain , The Economic History Review , 1987 , vol. 40 (pg. 561 - 70 ) . , Social Mobility in the 19th and 20th Centuries: Europe and America in Comparative Perspective , 1986 New York St. Martin's Press . , The Unbound Prometheus: Technological Change and Industrial Development in Western Europe, 1750 to the Present , 2003 Cambridge Cambridge University Press . , Wages and Earnings of the Working Classes , 1885 London J. Murray Report to Sir Arthur Bass, M.P . Unequal English wealth since 1670 , Journal of Political Economy , 1986 , vol. 94 (pg. 1127 - 62 ) . When did inequality rise in Britain and America? , Journal of Income Distribution , 2000 , vol. 9 (pg. 11 - 25 ) . , Growing Public: Social Spending and Economic Growth Since the Eighteenth Century. Further Evidence, Vol. 2 , 2004 Cambridge Cambridge University Press . Rural-urban migration and socioeconomic mobility in Victorian Britain , Journal of Economic History , 2005 , vol. 65 (pg. 1 - 35 ) , . Intergenerational occupational mobility in Britain and the U.S. since 1850 , American Economic Review Forthcoming . , . How open was nineteenth-century British society? Social mobility and equality of opportunity, 1839–1914 , Building European Society: Occupational Change and Social Mobility in Europe, 1840–1940 , 1993 Manchester Manchester University Press (pg. 18 - 39 ) . , Social Mobility in Nineteenth- and Early Twentieth-Century England , 1999 New York St. Martin's Press , . , Building European Society: Occupational Change and Social Mobility in Europe, 1840–1940 , 1993 Manchester Manchester University Press . , Principles of Political Economy , 1904 London Longmans, Green, and Co . , . “Inequalities which every one may remove”: occupational recruitment, endogamy, and the homogeneity of social origins in Victorian England , Building European Society: Occupational Change and Social Mobility in Europe, 1840–1940 , 1993 Manchester Manchester University Press (pg. 140 - 64 ) , , . , Origins of the Modern Career , 2004 Aldershot and Burlington, VT, USA Ashgate . , The Rise and Decline of Nations: Economic Growth, Stagflation and Social Rigidities , 1982 New Haven Yale University Press OPCS , Classification of Occupations , 1970 London HMSO . , Self-Help , 2002 Oxford Oxford University Press . , . Intergenerational mobility in the labor market , Handbook of Labor Economics , 1999 New York Elsevier (pg. 1277 - 3630 ) . Cross-country differences in intergenerational earnings mobility , Journal of Economic Perspectives , 2002 , vol. 16 (pg. 59 - 66 ) . . A model of intergenerational mobility variation over time and place , Generational Income Mobility in North America and Europe , 2004 Cambridge Cambridge University Press (pg. 38 - 47 ) . , Poverty and Progress; Social Mobility in a Nineteenth-Century City , 1964 Cambridge Harvard University Press . , The Other Bostonians , 1973 Cambridge Harvard University Press . , Literacy and Popular Culture: England 1750–1914 , 1989 Cambridge Cambridge University Press . Earnings inequality in nineteenth-century Britain , Journal of Economic History , 1980 , vol. 40 (pg. 457 - 75 ) . The structure of pay in Britain, 1710–1911 , Research in Economic History , 1982 , vol. 7 (pg. 1 - 54 ) . , Did British Capitalism Breed Inequality? , 1985 Boston Allen and Unwin , . , The Population History of England, 1541–1871 , 1989 Cambridge Cambridge University Press 1 Solon (1999, 2002) and Black and Devereux (2011) are helpful overviews of the large literature on intergenerational economic mobility. Corak (2006) provides an international comparison of recent results. For results on social class mobility in Britain, the seminal study is Goldthorpe (1987). For earnings mobility in Britain, Dearden et al. (1997) find a father–son earnings elasticity of between 0.4 and 0.6, depending on specification, leading them to characterize the extent of intergenerational earnings mobility as “limited.” Studies over a longer time period have been limited to the postwar period. Blanden et al. (2004) show that earnings mobility in Britain is low relative to other advanced countries and declined from the 1960s to the 1980s. Aaronson and Mazumder (2008) find that mobility in the USA increased from 1950 to 1980, but declined sharply from 1980 to 2000. 3 It should be noted that the analysis in this paper includes only England and Wales, not Scotland. The title is, therefore, not entirely accurate. The judgment here is that omitting the Scottish from the collective “British” is preferable to including the Welsh in the collective “English.” The terms “Britain” and “British” are used here, loosely, to refer to England and Wales. 4 For a concise survey of this literature, see Miles (1999, pp. 8–13). 6 The records were originally sampled by David Vincent, who gathered information on approximately 1,000 marriages from each of 10 Registration Districts at 5-year intervals from 1839 to 1914. The districts were selected to include a broad cross section of economic activity; they are Dudley, Sheffield, Macclesfield, Stoke-on-Trent, Lichfield, Bethnal Green, Wokingham, Nuneaton, Cleobury Mortimer, and Samford. The sample includes a total of 10,835 marriages. For more details on the marriage registry data, see Vincent (1989) and Miles (1999). 7 This is true insofar as the goal is to measure total mobility from father to son when both have reached a point of what Goldthorpe (1987, pp. 52–53) calls “occupational maturity,” that is, the point at which one would expect “a marked falling-off in the probability of job changes which involve major shifts of occupational level.” Indeed, this is the sense in which intergenerational mobility is most commonly understood. A related but separate question concerns the impact of father's occupation on son's first occupation, which is further discussed in Section 3. 8 The terms “intragenerational mobility” and “career mobility” are often taken to be synonymous, as they are here. This is perhaps an abuse of terminology, as the concept of a “career” implies more than a simple “occupation.” 9 Or for other countries in the nineteenth century for that matter. Referring to the difficulty of quantifying historical rates of career mobility, Miles (1999, p. 15) writes “most attention will be devoted to this aspect of mobility (intergenerational mobility), although this is partly because substantial and standardized material detailing career mobility is much harder to come by.” 10 The 1851 and 1881 datasets are available from the UK Data Archive, as studies number 1316 and 4177, respectively. 12 To test the robustness of the matching procedure, the intergenerational mobility analysis was repeated using only the individuals who were matched exactly between 1851 and 1881—in other words, those individuals who reported precisely the same name in both censuses and whose reported age in 1881 was exactly 30 years greater than their reported age in 1851. The mobility pattern for this group is essentially the same as the results reported in Table 2 for the data as a whole: total mobility is 49 percent, upward mobility 27 percent, and downward mobility 22 percent. The Altham G2 test statistic, described in Section 3, shows no statistically significant difference in mobility between the exact-match subset and the whole dataset. 13 Some typical occupations are Class I—solicitor, accountant; Class II—farmer, carpenter (employer); Class III—carpenter (not employer), butcher (not employer), skilled in manufacturing; Class IV—agricultural laborer, wool comber; Class V—general laborer, porter. 14 These numbers are derived from a modification of Williamson's wage estimates that is discussed below in Section 3 of this paper. 15 Williamson's pay estimates have come in for some harsh criticism; see for example Jackson (1987) and Feinstein (1988). However, this criticism is aimed almost exclusively at his estimates of white-collar pay. His pay estimates for the Class III skilled occupations that are used here are generally accepted. 16 For the linked census data, agricultural laborers have been recoded from class IV to class V, in accordance with the modification to Armstrong's scheme employed by Miles. 18 For more details on the data and the study itself, see Goldthorpe (1987). The data are available from the UK Data Archive as study number 1097. 20 OPCS (1970) contains the information necessary for the reclassification. 21 The same is true for intragenerational mobility, although the results are not shown. 22 The OMS questionnaire asked each individual to report income and occupation, which allows income by occupation and by occupational class to be calculated from the data. 23 Another advantage is the inclusion of some wage information for white-collar workers. Although the previously mentioned professional occupations are problematic, there are a handful of occupations included with which to derive an average wage for classes I and II. Other viable sources for occupational wages from one specific time period, such as Leone Levi's 1885 report, lack any wage information for the professional occupations (Levi 1885). 24 Statistical tests of the equality of the estimates of β across time periods indicate that the 1851/81 estimate is significantly different from the estimates for 1881/1901 and 1972, but the difference between the 1881/1901 and 1972 estimates is not statistically significant. AppendixTable A1. Summary of mobility data
Table A1. Summary of mobility data
Table A2. Example of linked census data, 1851–1881
Note: Three linked individuals who were used to measure intergenerational mobility are shown in italics. To enhance readability, some members of each household are not shown. Table A2. Example of linked census data, 1851–1881
Note: Three linked individuals who were used to measure intergenerational mobility are shown in italics. To enhance readability, some members of each household are not shown. Table A3. Intragenerational mobility, married males 1851–1881: column percentages (mobility rates)
Note: Married males younger than thirty-five in 1851. Table A3. Intragenerational mobility, married males 1851–1881: column percentages (mobility rates)
Note: Married males younger than thirty-five in 1851. Table A4. Intergenerational mobility, 1851–1881: wage mobility within class III
Notes: Sons aged 0–19 in 1851. All sons with Class III occupations in 1881; all fathers with Class III occupations in 1851. Wages are real annual wages calculated at 1851 prices. Wage information was taken from Williamson (1980, 1982). Table A4. Intergenerational mobility, 1851–1881: wage mobility within class III
Notes: Sons aged 0–19 in 1851. All sons with Class III occupations in 1881; all fathers with Class III occupations in 1851. Wages are real annual wages calculated at 1851 prices. Wage information was taken from Williamson (1980, 1982). © European Historical Economics Society 2013 © European Historical Economics Society 2013 Why was the Interstate Commerce Commission established quizlet?The Interstate Commerce Commission was established in 1887 to: ensure that railroads charged farmers and merchants reasonable and fair rates.
What did Henry Grady advocate quizlet?What did Henry Grady advocate? He advocated diversifying the economy and expanding industrial production in the South.
Which of the following was one of the secrets of John D Rockefeller's success?Which of the following was one of the secrets of John D. Rockefeller's success? He paid attention to the minutest details.
Where was baseball the most popular CH 19?Where was baseball the most popular? in urban areas with large working-class populations.
|
