full version ( ppt ) - Institute for Fiscal Studies

Report
Children’s outcomes and family
background
Claire Crawford
© Institute for Fiscal Studies
Introduction
• UK has relatively low intergenerational mobility
– Correlation between parents and children’s income is relatively high
– Intergenerational elasticity of 0.29 for those born in Britain in 1970
(Blanden et al, 2005)
– Circumstances into which you are born heavily influence future income
• Government would like to improve life chances of children in
poverty
– Child Poverty Act (2010)
– Review on Poverty and Life Chances
© Institute for Fiscal Studies
How can we improve life chances?
• If the link between income across generations is causal, then
increasing parents’ incomes today should lead to higher income
for their children in future
• Of course parental income may not be the only factor that is
causally related to children’s future income (or well-being)
– Other family background characteristics (e.g. Parents’ education,
marital status)
– Other factors (e.g. Health) – not dealt with today
© Institute for Fiscal Studies
An aside on causality . . .
• What do we mean by causal?
– Certainty that changing a particular factor of interest causes (rather
than is simply correlated with) a change in the outcome of interest
• Important distinction, because we really only want to base policy
decisions on causal (rather than correlational) relationships
© Institute for Fiscal Studies
How can we improve life chances?
• If the link between income across generations is causal, then
increasing parents’ incomes today should lead to higher income
for their children in future
• Of course parental income may not be the only family
background characteristic that is causally related to children’s
future income
– Others include parents’ education and marital status
© Institute for Fiscal Studies
Evidence on causal relationships
• Can be difficult to find evidence of direct causal relationships
between (e.g.) parent and child income, not least because of
need for long time lags between observations
• Can instead piece together evidence in two stages:
– Establish causal links between outcomes in childhood and later in
life
– Establish causal links between parental income (and other family
background characteristics) and child outcomes
© Institute for Fiscal Studies
Link between outcomes in childhood and later
life
• Evidence of causal relationship between children’s educational
attainment and their future income
– e.g. Blundell et al (1999) suggest that the gross rate of return to an
additional year’s education in the UK is 5-10%
• Also some evidence of causal link between childhood cognitive
and non-cognitive skills and a range of adult outcomes
– e.g. Heckman et al (2006) for the US; Carneiro et al (2008) for the
UK
• Suggests improving educational attainment and skills amongst
poor children is key to improving future labour market outcomes
– Other outcomes also relevant for wider well-being, e.g. health
© Institute for Fiscal Studies
Link between family background and child
outcomes: focus on parental income
• Some evidence of causal relationship between parental income
and children’s educational attainment
– e.g. Blanden & Gregg (2004) suggest that a one third fall in
household income (around £7,000) reduces the probability of getting
a degree by around 5 percentage points (ppts)
– e.g. Chevalier et al (2005) suggest that a doubling of father’s income
increases the likelihood of post-compulsory education by 14 ppts
– Suggests that policies which focus on increasing parents’ income
are likely to improve children’s life chances - although such sizeable
income changes may be beyond the scope of policymakers to
provide
© Institute for Fiscal Studies
Link between family background and child
outcomes: focus on parental education
• Also of causal link between parents’ and children’s education
– e.g. Chevalier (2004) suggests that each additional year of parental
education increases the probability of staying on in post-compulsory
education by up to 8 ppts (although estimates are insignificant)
– Chowdry et al (2008) also suggest that parental education increases
GCSE attainment (although again estimates are insignificant)
– Suggests that improving educational attainment amongst today’s
children will also have benefits for the next generation
© Institute for Fiscal Studies
Link between family background and child
outcomes: focus on marital status
• What about other family background characteristics?
• Evidence from UK and elsewhere that children born to married
parents have better cognitive and behavioural outcomes than
children born to cohabiting parents
– Conservatives’ proposal to support marriage through the tax system
presumably based on such evidence
• But do these gaps reflect a causal effect of marriage on child
outcomes? Or do they simply reflect the fact that different sorts
of people choose to get married (selection effect)?
– Recent IFS research tries to shed light on this issue . . .
© Institute for Fiscal Studies
Difference in outcomes between children of
married and cohabiting parents at birth
Standard deviations (mean = 0)
0.2
0.1
0
BAS (age 3)
BAS (age 5)
SDQ (age3)
-0.1
-0.2
-0.3
-0.4
Cohabiting
© Institute for Fiscal Studies
Married
Difference
SDQ (age 5)
Aside: how does this gap compare to others?
Cognitive development at age 3
Gap (in standard deviations)
1
0.75
0.5
0.25
0
Mother high White vs non- Top vs bottom
Father
Lone parents
vs low
white
income
professional vs any couple
education
quintile
vs routine
occupation
© Institute for Fiscal Studies
Married vs
cohabiting
Aside: how does this gap compare to others?
Social and emotional development at age 3
Gap (in standard deviations)
1
0.75
0.5
0.25
0
Education
© Institute for Fiscal Studies
Income
Occupation
Lone
parenthood
Married vs
cohabiting
White vs nonwhite
Link between family background and child
outcomes: focus on marital status
• What about other family background characteristics?
• Evidence from UK and elsewhere that children born to married
parents have better cognitive and behavioural outcomes than
children born to cohabiting parents
• Conservatives’ proposal to support marriage through the tax
system presumably based on such evidence
• But do these gaps reflect a causal effect of marriage on child
outcomes? Or do they simply reflect the fact that different sorts
of people choose to get married (selection effect)?
– Recent IFS research tries to shed light on this issue . . .
© Institute for Fiscal Studies
Aim of research
• To provide a best estimate of the causal impact of marriage on
child outcomes by eliminating that part of the gap due to
selection
– i.e. take account of the fact that people who choose to get married
are different from those who do not
• Interpret the remaining gap as the causal effect of marriage
• Need to strike a careful balance in terms of controls:
– “Over-control” and you risk under-estimating the effect of marriage
– “Under-control” and you risk over-estimating the effect of marriage
© Institute for Fiscal Studies
Data
• Millennium Cohort Study
• We use a sample of around 10,000 children, born to married or
cohabiting couples (i.e. we exclude lone parents)
• Marital status measured at birth
– 70% married; 30% cohabiting
• Outcomes:
– Cognitive development measured using vocabulary component of
British Ability Scales (BAS) at ages 3 and 5
– Social and behavioural development measured using motherreported Strengths and Difficulties Questionnaire (SDQ) at ages 3
and 5
© Institute for Fiscal Studies
Controls
• Three key groups of variables:
Group 1
Group 2
Group 3
Reflect sort of person
that chooses to get
married
Very likely
Likely
Possible
Affected by marriage
Unlikely
Possible
Likely
Examples...
Ethnicity;
religion
Education;
occupation;
housing tenure;
income
Relationship
stability;
parenting
practices
• We believe that controlling for groups 1 and 2 is the right balance
to strike to identify the causal effect of marriage. But debatable . .
.
© Institute for Fiscal Studies
Who cohabits rather than marries?
• Cohabiting parents are more likely than married parents to:
– Be White or Black Caribbean
– Be of no religion
– Be low qualified
– Be home renters rather than homeowners
Groups 1 and 2
– Be teenagers at birth of first child
– Have lived together for short time
– Report that the pregnancy was unplanned
– Exhibit lower relationship quality (at 9 months)
• They are also more likely to:
– Have poorer maternal mental health (at 9 months)
– Have lower paternal involvement with baby (at 9 months)
3
– Be less likely to set regular bedtimes (at age 3)
© Institute for Fiscal Studies
Group
Explaining difference in outcomes between
children born to married vs. cohabiting parents
Outcome
A
B
C
D
E
F
BAS (age 3)
-0.101**
-0.155**
0.008
0.065
0.041
0.048
SDQ (age 3)
-0.325**
-0.316**
-0.187**
-0.129**
-0.068
-0.039
BAS (age 5)
-0.192**
-0.229**
-0.056
-0.017
-0.049
-0.044
SDQ (age 5)
-0.301**
-0.291**
-0.176**
-0.125**
-0.086*
-0.054
A controls for the child’s month and year of birth
B also controls for mother’s ethnicity, immigration status and religion
C also controls for education and socio-economic classification of the parents
D also controls for household income, tenure and work at 9 months
E also controls for family structure at 9 months
F also controls for relationship quality at 9 months
© Institute for Fiscal Studies
Summary: impact of marital status on child
outcomes
• Small gap in cognitive ability at ages 3 and 5, largely explained
by the fact that, compared to married parents, cohabiting
parents:
– Have lower education
– Have lower occupational status
– Have lower income
– Are more likely to live in social housing
• Larger gap in social and emotional development at ages 3 and
5, largely explained by the fact that cohabiting parents:
– Have lower education
– Have lower socio-economic status
– Are more likely to have unplanned pregnancies
– Report lower relationship quality when their child is 9 months old
© Institute for Fiscal Studies
Conclusions on marital status
• Differences in outcomes between children born to married and
cohabiting couples largely reflect differential selection into
marriage, rather than a causal effect of marriage itself
– Otherwise marriage needs to lead to very significant improvements
in parents’ socio-economic status and relationship quality
• Suggests that providing a tax incentive to encourage more
parents to get married is unlikely to significantly improve child
outcomes
© Institute for Fiscal Studies
Broader conclusions
• Policies that improve educational attainment amongst poor
children are likely to have long-term pay-offs in terms of
increasing life chances amongst this generation and the next
• Amongst the family background characteristics I have
considered:
– Increasing parental income or education may help achieve these
aims
– Encouraging more parents to marry probably will not
• Of course, there are other ways to raise educational attainment
amongst poor children as well
– e.g. raising school quality; improving attitudes and behaviours
• But causal evidence on the latter much less clear
– Clear need for well-designed policy experiments
© Institute for Fiscal Studies

similar documents