Download presentation
Presentation is loading. Please wait.
Published byClay Scoggin Modified over 10 years ago
1
Social Mobility Within and Across Generations in Britain Since 1851 Jason Long Department of Economics Colby College November 2008
2
Social Mobility in 19 th Century Britain Why Does Mobility Matter? 1.Fundamental to our understanding of economic equality and “fairness” of society. Importance of Distribution is obvious, and well-studied – income, earnings, wealth, etc. But Mobility informs our understanding of a given distribution: two societies with identical earnings distributions but different mobility regimes are not equally equal (Stokey, 1996). Distribution→ Equality of Outcome Mobility→ Equality of Opportunity
3
Why Does Mobility Matter? 2.Has implications for economic efficiency. Social/Class rigidities can impeed efficient allocation of labor If educational institutions and labor market fail to sort, select, and promote talent, allocation of human capital in economy is suboptimal Landes (1969) very critical of late 19 th century English economy in this regard, especially with respect to education 3.Can be crucially important for long-run path of public policy: Current perceptions of mobility and equality of opportunity → Preferences for taxation/redistribution → Continuting perceptions of mobility/equality Piketty (QJE 1995) Bénabou and Ok (QJE 2001) Bénabou and Tirole (NBER 2005)
4
Equality in 19 th Century Britain Wage and Wealth Distribution Subject of economic inequality and “Kuznets hypothesis” has received much attention: Williamson (1980, 1982, 1985): Pay ratios show Kuznets curve – stability from late 18 th c until early 19 th, rising inequality until mid-century, leveling up to WWI Feinstein (1988): No trend during 19 th century Lindert (1986, 2000, 2002): Real inequality increased earlier than previously thought, from 1740 – 1810. Wealth inequality greater before 1914 than since 1950.
5
Equality in 19 th Century England Social Mobility: Empirical Evidence Major work is Miles, Social Mobility in 19 th and Early 20 th Century England (1999). Also Mitch (1993) and Miles (1993). 68% of sons in same occupational class as father from 1839-54, falling to 53% by 1899-1914. “In terms of its inhabitants’ relative life chances, [Victorian and Edwardian England was] a profoundly unequal society.” 20 th Century Results: Goldthorpe (1980):51% in 1972 in same class as father 58% in same class as first full-time job Mobility trendless in 20 th century: “Constant social fluidity” Baines, Johnson (1999): Higher working class mobility (50%) in interwar London than in England from 1899-1914 (40%) Dearden et al (1997): Father/Son earnings elasticity between 0.4 and 0.6
6
Studying Mobility in 19 th Century England Marriage Registry Data All previous national-level studies have relied on data from marriage registries. 1836 Registration of Births, Deaths, and Marriages Act: church registers must record occupation of bride, groom, and parents Advantages:(1) Unique in recording fathers’, sons’ occupations (2) Signitures provide proxy for literacy Disadvantages ◦Excludes non-marrying population (10% of 45-year old males) ◦Includes only Anglican ceremonies (by 1914, over 40% of marriages were non-Anglican) ◦“Snapshot problem”: Father and son at point in time Does not control for stage of life cycle ◦Cannot observe intra-generational mobility at all
7
Research Questions 1.What was the rate of intergenerational social mobility in nineteenth- century Britain? How does controlling for life-cycle effects change what we know about mobility? 2.How prevalent was intra-generational mobility? How does it compare to mobility across generations? 3.How does mobility in nineteenth-century Britain compare with Britain in the twentieth century? What is the long-run trend in mobility? 4.What were the determinants of mobility for the individual? How important were human capital investments such as schooling and geographic mobility? 5.How persistent was mobility across three generations?
8
New Approach to Measuring Mobility: Linked Census Data 2% Sample of 1851 Census Complete-Count 1881 Census 168,130 men in England and Wales All 12,640,000 men in the census 28,474 men in 1851 and 1881 16,829 sons in 1851 9,477 HH heads in 1851 20,269 sons in 1881 + Complete-Count 1901 Census + 8,677 sons in 1901 Match Criteria: Name (phonetic) Year of birth ( 5) County of birth Parish of birth
9
Example: William, William, and David Phillips
10
Linked Data: Three Generations, 1851 – 1901 End result: 54,218 males covering three generations from 1851 to 1901 Inter-generational mobility, 1851–1881 ◦12,647 father/son pairs where son < 20 years old in 1851 ◦Average age of father in 1851 = 41.5 years ◦Average age of son in 1881 = 38.0 years Intra-generational mobility, 1851–1881 ◦7,790 males aged 20-35 in 1851, 50-65 in 1881 Inter-generational mobility, 1881–1901 ◦4,071 father/son pairs where son between 10-19 years old in 1881 ◦Average age of father in 1881 = 46.7 years ◦Average age of son in 1901 = 33.9 years Mobility over three generations, 1851–1881–1901 ◦5,763 grandfather/father/son sets ◦Average age of grandfather in 1851 = 45.8 years ◦Average age of father in 1881 = 45.3 years ◦Average age of son in 1901 = 32.1 years
11
Advantages of Linked Census Data for Studying Mobility Controls for life cycle: “destination to destination” Allows intra-generational mobility to be observed Three generations Nationally Representative Large: High cell counts Household background information: ◦Siblings, Birth order ◦Mother’s status ◦Servants ◦School attendance ◦Place of residence Reveals geographic mobility along with occupational
12
Measuring Mobility Focus here is on occupational/class mobility, rather than on earnings mobility (more common in the economics literature), for three reasons: It’s all there is. Census doesn’t include any information on wage, wealth, or consumption. Subject to fewer transitory shocks than earnings, one of the principal empirical difficulties with modern earnings elasticity studies. Captures more dimensions of an individual’s experience that may be related to views of mobility: ◦standing in the community (prestige) ◦control at work (supervisory v. non-supervisory) ◦ability to control hours, importance of risk (self-employed v. others) ◦manual v. non-manual ◦place of work (factory/office v. home/farm)
13
Making Use of Occupation Option 1: Occupational Classification (Armstrong, 1972) Class I.Professional etc., occupations accountant, solicitor, surgeon, large employer Class II.Intermediate occupations bookkeeper, manager, farm foreman, craftsman (employer) Class III.Skilled occupations blacksmith, nurseryman, weaver, craftsman (non-employer) Class IV.Semiskilled occupations agricultural labourer, flax dresser, rat destroyer Class V.Unskilled occupations general labourer, rag & paper collector, bone sorter 3.5% in 1881 15.5% 50.8% 17.8% 12.4%
14
Making Use of Occupation Option 2: Imputed Earnings
15
Intergenerational Social Mobility, 1851-1881 Sons aged 0-19 in 1851, 30-49 in 1881 MOBILITY: Total:50.11% Up:26.78% Down:23.33%
16
Social Mobility 1851-1881 Intra- and Inter-Generational Mobility MOBILITY: Total:43.59% Up:25.38% Down:18.22% MOBILITY: Total:50.11% Up:26.78% Down:23.33%
17
Intergenerational Mobility: Controlling for Life Cycle Effects Linked Census Data vs Marriage Registry Data MOBILITY: Total:47.54% Up:27.82% Down:19.71% MOBILITY: Total:34.80% Up:17.72% Down:17.08%
18
Comparing Mobility Across Tables Need a single metric that summarizes difference in mobility across two tables is not affected by differences in occupation structure across tables can be tested for statistical significance Use cross-product ratio(s). For matrix M, Invariant to different occupational distributions Multiplying rows, columns by constants does not change underlying row-column association Multiply row 1 by r 1, row 2 by r 2, column 1 by s 1, column 2 by s 2, immobility mobility
19
Comparing Mobility Across Tables Altham (1970): For two r s tables, measures how far the association between rows and columns in table P departs from the association between rows and columns in table Q. A simple likelihood-ratio 2 statistic G 2 tests whether the matrix with elements θ ij =log(p ij /q ij ) is independent If d(P,Q) > 0 and d(P,1) > d(Q,1), greater mobility in Q (mobility is closer in Q than in P to what we would observe under independence of rows and columns.)
20
Intergenerational Mobility: Controlling for Life Cycle Effects Linked Census Data vs Marriage Registry Data MOBILITY: Total:47.54% MOBILITY: Total:34.80% ALTHAM TESTS: d(P,1)46.26 G2G2 2885.98 prob [d(P,1)=0]0 d(Q,1)62.56 G2G2 1680.94 prob [d(Q,1)=0]0 d(P,Q)26.26 G2G2 267.17 prob [d(P,Q)=0]0
21
Research Questions 1.What was the rate of intergenerational social mobility in nineteenth- century Britain, controlling for life cycle? Higher than previously believed: Total mobility= 48% versus 35%, Upward mobility= 28% versus 18% 2.How prevalent was intra-generational mobility? Mobility within work-life was common: 44% changed class from 20s to 50s, 25% of them moving up 3. How does mobility in nineteenth-century Britain compare with Britain in the twentieth century? What is the long-run trend in mobility?
22
Comparing Mobility in Two Economies Simple two-generation human capital model generates clear predictions. Solon (1999, 2004), Becker and Tomes (1979, 1986):
23
Comparing Mobility in Two Economies Simple two-generation human capital model generates clear predictions. Solon (1999, 2004), Becker and Tomes (1979, 1986):
24
Comparing Mobility in Two Economies Simple two-generation human capital model generates clear predictions. Solon (1999, 2004), Becker and Tomes (1979, 1986):
25
Comparing Mobility in Two Economies Simple two-generation human capital model generates clear predictions. Solon (1999, 2004), Becker and Tomes (1979, 1986):
26
Comparing Mobility in Two Economies Simple two-generation human capital model generates clear predictions. Solon (1999, 2004), Becker and Tomes (1979, 1986):
27
Comparing Mobility in Two Economies Simple two-generation human capital model generates clear predictions. Solon (1999, 2004), Becker and Tomes (1979, 1986):
28
Comparing Mobility in Two Economies Simple two-generation human capital model generates clear predictions. Solon (1999, 2004), Becker and Tomes (1979, 1986):
29
Comparing Mobility in Two Economies Simple two-generation human capital model generates clear predictions. Solon (1999, 2004), Becker and Tomes (1979, 1986): with the standard earnings elasticity regression function:
30
Comparing Mobility in Two Economies Simple two-generation human capital model generates clear predictions. Solon (1999, 2004), Becker and Tomes (1979, 1986): with the standard earnings elasticity regression function: the model above implies
31
Comparing Mobility in Two Economies This implies that the intergenerational transmission of earnings will be greater (i.e. that mobility will be lower) when Heritability of “intrinsic” human capital ( λ ) is greater Human capital investment is more productive ( θ greater ) Earnings return to human capital ( p ) is greater Public investment in children’s human capital is less progressive ( γ less positive) Grawe & Mulligan (2002) and Han &Mulligan (2001) also show that Mobility is lower where capital markets function poorly Mobility is lower with more variance in ability
32
Intergenerational Mobility from 1851 to the Present 20 th century data: Oxford Mobility Study (1972) 10,309 males surveyed Ages 20 – 64 (mean = 41.1 years) Data on occupations: ◦Own occupation at time of survey ◦Father’s occupation when respondent aged 14 ◦Own first full-time occupation ◦Own occupation 10 years after first Occupations coded into same five-tiered scheme (based on Registrar General’s 1951 occupational classification system)
33
Intergenerational Mobility from 1851 to the Present Linked Census Data vs Oxford Mobility Study (1972) MOBILITY: Total:50.11% Up:26.78% Down:23.33% MOBILITY: Total:59.02% Up:35.52% Down:23.50%
34
Intergenerational Mobility from 1851 to the Present Linked Census Data vs Oxford Mobility Study (1972) MOBILITY: Total:50.11% MOBILITY: Total:59.02% ALTHAM TESTS: d(P,1)48.15 G2G2 2943.77 prob [d(P,1)=0]0 d(Q,1)33.26 G2G2 337.23 prob [d(Q,1)=0]0 d(P,Q)21.16 G2G2 247.37 prob [d(P,Q)=0]0
35
Mobility Trend, 1880-1970 Degree of Association between Fathers’ and Sons’ Occupations
36
Mobility Trends in Britain and the U.S., 1880-1970 Degree of Association between Fathers’ and Sons’ Occupations Sources: Long, 2008; Long and Ferrie, 2008
37
Research Questions 1.What was the rate of intergenerational social mobility in nineteenth- century Britain, controlling for life cycle? 2.How prevalent was intra-generational mobility? 3.How does mobility in nineteenth-century Britain compare with other countries then and with Britain in the twentieth century? Britain became more mobile from 1880 to 1970 Contrasts with Erickson and Goldthorpe’s finding of “constant flux” for many countries since WWII Consistent with Miles finding of upward trend from 1839 to 1914 Britain significantly less mobile in 19 th century than U.S. But, trends moving in opposite directions: mobility in the U.S. has declined dramatically since the 19 th century.
38
19 th Century British Mobility in Perspective Social mobility was significantly greater than previously believed Intragenerational mobility was substantial Mobility in general, however, was lower than in 1970, and lower than in the 19 th century U.S. How do we explain the differences?
39
Comparing Mobility in Two Economies Intergenerational transmission of earnings will be greater (i.e. mobility will be lower) when Heritability of “intrinsic” human capital ( λ ) is greater Human capital investment is more productive ( θ greater ) Earnings return to human capital ( p ) is greater Public investment in children’s human capital is less progressive ( γ less positive) Capital markets function poorly There is greater variance in ability
40
The (Belated) Rise of Schooling in England Education Act of 1870 establishes school boards; mandatory, government-funded primary education Education Act of 1880: Set minimum leaving age to 10, heavily restricted to 13 Leaving age periodically raised thereafter, to age 14 by 1900
41
Research Questions 1.What was the rate of intergenerational social mobility in nineteenth- century Britain, controlling for life cycle? 2.How prevalent was intra-generational mobility? 3.How does mobility in nineteenth-century Britain compare with other countries then and with Britain in the twentieth century? 4.What were the determinants of mobility for the individual? How important were childhood human capital investments and geographic mobility?
42
Micro-Level Analysis What were the empirically important determinants of mobility? Simple reduced-form estimation of return to human capital and location characteristics: where h is a vector of human capital indicators p is earnings return to human capital l is a vector of location characteristics
43
Micro-Level Analysis What were the empirically important determinants of mobility? For occupational mobility: where y * is true, unobserved job quality y is observed job class and p, p u > 0, p d 0
47
Accounting for Endogeneity Structural mobility equations: where y is imputed wage h 1 is an endogenous human capital indicator h 2 is a vector of exogenous human capital indicators l is a vector of exogenous location characteristics z is a vector of variables that determine the choice of h 1 Identification requires exclusion restriction(s): variable(s) in z not present in h 2, l.
48
Endogenous Migration Decision (Internal) migration choice endogenous. Expected wage/mobility gains influence location choice Potential selectivity bias Structural Model of Endogenous Migration Choice: Move/stay choice and treatment effect of migration jointly estimated by maximum likelihood
49
Identification Identification comes from two sources: Nonlinearity of model m i appears in wage equation, while m i * appears in migration decision equation Variable included in z but excluded from x Not in town of birth in 1851 F = 371.51
51
Endogenous School Choice School attendance may not be exogenous. Parents consider net benefit to child of schooling. Potential “ability bias”. Structural Model of Endogenous School Choice (Heckman, 1979, 1990; Harmon & Walker, 1995; many others) Selection of school attendance over non-attendance and treatment effect of schooling jointly estimated by maximum likelihood
52
Identification Identification comes from two sources: Nonlinearity of model – s i appears in wage equation, while s i * appears in school decision equation Variables included in z but excluded from x ◦Schools / 1,000 residents (county) F = 0.23 ◦Older sister in household ◦Number of younger siblings ◦Father's age ◦Percent sons F = 2.63 ◦School attendance of siblings F = 78.02
54
Research Questions 1.What was the rate of intergenerational social mobility in nineteenth- century Britain, controlling for life cycle? 2.How prevalent was intra-generational mobility? 3.How does mobility in nineteenth-century Britain compare with other countries then and with Britain in the twentieth century? 4.What were the determinants of mobility for the individual? How important were childhood human capital investments and geographic mobility? Childhood investments were clearly productive: schooling, birth order, and numeracy mattered, as did servants “Locational investments” were also productive: county movers were 56% more likely to be upwardly mobile (34 vs 22 percent)
55
Research Questions 1.What was the rate of intergenerational social mobility in nineteenth- century Britain, controlling for life cycle? 2.How prevalent was intra-generational mobility? 3.How does mobility in nineteenth-century Britain compare with other countries then and with Britain in the twentieth century? 4.What were the determinants of mobility for the individual? 5.How persistent was mobility across three generations?
56
Example: William, William, and David Phillips
57
Mobility Across Three Generations Intergenerational Mobility 1881-1901, By Father's IG Mobility 1851-1881 Fathers in 1881 in same class as their father in 1851 Fathers in 1881 in higher class than their father in 1851 Fathers in 1881 in lower class than their father in 1851
58
Mobility Across Three Generations Intergenerational Mobility 1881-1901, By Father's IG Mobility 1851-1881 MOBILITY: Total:42.49% Up:22.14% Down:20.35% MOBILITY: Total:48.96% Up:12.47% Down:36.49% MOBILITY: Total:60.99% Up:52.23% Down:8.76%
59
Wage Elasticities Across Three Generations Simplest test of “Grandfather Effect”: log y i,1901 = β 0 + β 1 log y i,1881 + β 2 log y i,1851 + ε i where y i,1901 is individual’s imputed wage as adult in 1901 y i,1881 is individual’s father’s imputed wage as adult in 1881 y i,1851 is individual’s grandfather’s imputed wage as adult in 1851 Results: β 1 = 0.283 (s.e. 0.012) β 2 = 0.054 (s.e. 0.011)
60
Research Questions 1.What was the rate of intergenerational social mobility in nineteenth- century Britain, controlling for life cycle? 2.How prevalent was intra-generational mobility? 3.How does mobility in nineteenth-century Britain compare with other countries then and with Britain in the twentieth century? 4.What were the determinants of mobility for the individual? 5.How persistent was mobility across three generations? Generational effects persistent: Grandfathers mattered. Why? Standard models do not include direct grandparent effects…only indirect through parents, who are the ones to invest in child’s human capital. ◦Persistent wealth effects ◦Unobserved variance in job quality within broad class ◦Persistent class barriers (social networks, stigma, etc) that operate across several, perhaps many, generations
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.