Social class and income intergenerational mobility

1
Centre for Market and Public Organisation
Intergenerational Mobility The Next
Generation Based on work by Jo Blanden, Paul
Gregg and Lindsey Macmillan, Longview Seminar
29th June 2007.
2
Introduction Background

• Most people are willing to accept wide
  inequalities if they are coupled with equality of
  opportunities The Economist (Oct 2006)
• This year I will be Talking More about Social
  Mobility David Cameron Today Programme, Radio 4
  (Jan 2007)

3
Introduction Background

• Intergenerational
• Income Mobility
• How closely related are incomes of parents and
  children?
• Social Mobility
• How closely related is the Social class of
  parents and children?
• Also other outcomes

4
Introduction Background

• Literature was mostly concerned with measurement,
  i.e. the strength of the correlation between
  income across generations.
• More recently comparisons across countries and
  across time have begun to emerge
• 2 substantive findings In terms of Income based
  mobility
• UK relatively immobile (along with US)
• Mobility in UK has declined between 1958 and 1970
  birth cohorts

5
Methodology
Intergenerational Income Mobility

6
Introduction Background

• Recent and On-going CMPO/CEE Research
• 1. Drivers of Mobility
• 2. Comparing Drivers of Mobility across cohorts
• 3. Projecting Mobility for Current generations

7
Plan of the Talk

• Consider the routes through which income persists
  for the 1970 (BCS) cohort. The objective is to
  understand the drivers of persistence.
• Analysis is restricted to sons at this stage.
• Make comparisons between the 1958 (NCDS) and 1970
  cohorts in an attempt to understand why
  intergenerational transmissions have
  strengthened.
• Extend analysis to earlier and later cohorts for
  within data restrictions

8
Drivers of Mobility Modelling approach
9
Data British Cohort Study (1)

• Parental income data available at ages 10 and 16,
  average these.
• Sons earnings at age 33.
• Cognitive tests at age 5 and 10.
• Mother reports on behaviour age 5.
• Teacher reports on behaviour and self-reported
  measures at age 10.
• Detailed education information including exam
  results.
• Work history records from age 30 enable the
  construction of number of months unemployed and
  out of labour force.

10
Data British Cohort Study (2)

• Cognitive tests
• Age 5 copying and english picture vocab test
• Age 10 reading, maths, British ability scale
• Non-cognitive measures
• Mum, age 5 neurotic, anti-social
• Teacher, age 10 application, clumsiness,
  extroversion, hyper-activity, anxious.
• Child, age 10 locus of control, self-confidence.
• Child, age 16 malaise.
• All cognitive and non-cognitive measures are
  normalised to mean 0, standard deviation 1.

11
Family income relationships - 1970
12
Earnings Equations at age 30
13
(No Transcript)
14
Data Cross cohort comparison

• Income is only available at age 16 in NCDS.
  Earnings are from age 33.
• Cognitive tests for reading, maths and general
  ability at 11, similar to BCS.
• Non-cognitive tests are different between the
  cohorts, use Bristol social adjustment scales for
  NCDS.
• unforthcoming, withdrawn, depressed, anxious for
  acceptance adults, hostile to adults, writing
  off adults, anxious for acceptance kids,
  hostility to kids, restless, inconsequential
  behaviour, misc.
• For both cohorts mother reports generate two
  measures from rutter scales at age 10,
  internalising and externalising.
• Concerns about attrition and non-response in both
  cohorts, no evidence that this is responsible for
  cross-cohort differences.

15
The change in intergenerational mobility
16
(No Transcript)
17
Policy implications

• Fall in mobility was explained by growing
  relationship between family income and
  non-cognitive skills, education and early
  unemployment.
• Not due to IQ or cognitive skills.
• 3 possible policy routes
• Close gap in non-cognitive skills (especially
  personal efficacy and concentration).
• Educational performance at age 16 and beyond.
• Help in early career (policies to avoid NEET).

18
Conclusions

• Overall nearly 90 of the increase in
  intergenerational income is explained using the
  first approach
• 20 of this rise is accounted for by the
  increasing correlation between income and O
  levels
• 35 through post-16 education (A-levels, degree
  etc)
• 30 through early LM attachment youth
  unemployment/NEETs
• None came through innate cognitive ability being
  more related to family background and ability
  became less important in predicting education
  attainment
• It is from within social class variation in
  income that the increase in persistence has
  occurred

19
2. Extending the Story

• Intergenerational Mobility has an ageing
  problem, to observe the full cycle takes at least
  30 years.
• But we can observe the childhood drivers of
  mobility evolving much quicker Family income
  relationship with test scores, education etc.
• And can observe returns of the most recent cohort
  available this can provide a prediction of
  mobility for the generation of children.
• Also the earlier 1946 cohort offers further
  insight into past intergenerational patterns

20
Four Cohorts

• Two Parts
• Family Income and educational attainment at age
  18 (58,70,75-80, 81-86 - BHPS) or at 21 (58,70,
  75-80)
• Family Income and Test scores (58,70,91/2)

21
Observable data for first stage analysis

• 1958 1970 1975-86 1991/2
• Family income v v v v
• Parental Education v v v
  v
• Childs Education v v v
  
• Cognitive scores v v v
• (at age 10/11)
• Self-esteem v v
  
• Locus of Control v v
• Application v v v
• Adult Earnings v v

22
Relationship between family income and Childs
Education
23
(No Transcript)
24
Returns to education NCDS, BCS and BHPS old
25
Summary

• Family Background and test scores show very
  similar patterns over time (46-70)
• Relationship between Family Income and
  Educational qualifications strengthened and then
  receded
• Returns to qualifications very stable 1991 to 2002

26
Some Assumptions

• Assume returns remain the same for next cohort
• Assume residual earnings and family income have
  similar relationship

27
Run on to Later BHPS Cohort
28
And Finally - ALSPAC
29
Conclusions 1

• Cross cohort comparisons are difficult for data
  comparability reasons
• Relationships between family background and
  cognitive test scores (age 10) are very stable
• Educational attainment appears to have became
  more socially graded but this maybe easing

30
Conclusions 2

• Making some assumptions about returns to
  education and the extent to which residual
  earnings are related to family background we can
  project future mobility patterns
• This suggests that for children born in early
  1980s (and left school around 2000), mobility had
  return to the pictures observed for the 1958
  children