Centre for Market and Public Organisation

1
Centre for Market and Public Organisation
The effects of shared parental childcare in the
first three years on childrens school readiness
Evidence from the ALSPAC cohort Elizabeth
Washbrook CMPO Seminar 12th October, 2006
2
Motivation

• The early years are a vitally important period in
  childrens development

The most efficient strategy for strengthening
the future workforce, both economically and
neurobiologically, and improving its quality of
life is to invest in the environments of
disadvantaged children during the early childhood
years. Knudsen, Heckman, Cameron and Shonkoff,
2006

• School readiness has been the focus of a large
  body of research (e.g. The Future of Children,
  Spring 2005)
• Impact on resources required of school and other
  agencies
• Association with illiteracy, teen parenthood,
  juvenile delinquency and poorer educational
  qualifications (Baydar et al., 2003 Rouse et
  al., 2005)
• Interest in the effects of early intervention
  programmes

3
Motivation

• Changing social roles of mothers and fathers
• Employment rates of mothers of children under 5
  up from 25 in the 1970s to 55 in 2006
• Fathers childcare as a main activity up from
  under 15 mins to over 2 hours a day over the same
  period
• ALSPAC gives a unique opportunity to study
• The roles of mothers and fathers in the early
  years for children born in the 1990s
• The effects of trends towards gender equality on
  childrens welfare
• Previous research has found
• Negative association between very early maternal
  employment and child outcomes (Gregg et al.,
  2005 Ruhm, 2004)
• Positive association between qualitative measures
  of fathers involvement and later outcomes
  (Flouri and Buchanan, 2004)
• Introduction paid paternity leave in April 2003
  good for kids?

4
Overview

• Definitions of shared parental childcare and
  school readiness
• Organising framework
• The relationship between parental endowments and
  shared parenting
• The relationship between child endowments and
  shared parenting
• Effects on school readiness and the role of
  selection bias
• Heterogeneity in the effects of shared parental
  care
• Testing some potential explanations
• Testing the age at which the effects emerge
• Conclusions

5
Findings Preview

• The effects of shared parental childcare in the
  first 3 years (relative to maternal-only parental
  care) depend crucially on
• The gender of the child
• The age of the child when paternal childcare took
  place
• The intensity of paternal childcare
• Boys who experience long hours of paternal
  childcare ( gt15 hours per week) when they are
  toddlers perform poorer on academic assessments
  at entry to school than boys experiencing
  maternal-only parental care
• This finding is robust but why?
• Children who experience moderate hours of
  paternal childcare (5-15 hours) have better
  behavioural outcomes at age 4, but only if this
  care is begun after infancy

6
Definitions of shared parental childcare and
school readiness

• Data ALSPAC cohort of children born in 1991/1992
• Sample selection 6010 households with biological
  parents co-resident continuously from pregnancy
  until at least age 4
• Parental childcare Considered shared if father
  regularly looked after the child without the
  mother present for at least 5 hours a week
• School readiness
• Academic Teacher-assessed Entry Assessment in
  first year of schooling (Language, Reading,
  Writing, Mathematics)
• Behaviour Mother-assessed SDQ at 47 months
  (Hyperactivity, Emotional symptoms, Conduct
  problems, Peer problems)
• Types of effect
• Boys and girls
• Moderate intensity (5-15 hours) and high
  intensity (more than 15 hours)
• Year 1 and Years 23

7
Organising framework

• Gary Beckers household production model

(1)
(2)
(3)
(4)
(5)
(6)
Clear distinction Exogenous market and
non-market endowments ( wM, wF, v, E ) Endogenous
choices of inputs of time and money ( tMi, tFi,
xi )
8
What kind of families use shared parental
childcare?
9
What kind of families use shared parental
childcare?
10
Parental endowments and shared parenting

• Parents socio-economic endowments
• Education, occupation, age, quality of services
  in local area, availability of other family
  carers, housing tenure, family size
• Parents non-market endowments
• Mental and physical health, attitudes to
  parenthood, locus of control, smoking, drinking,
  attitudes to education
• The theory of the intra-household division of
  labour implies shared parental care should be
  more common where
• The fathers market capital is lower (less
  educated, younger men in insecure low skilled
  occupations deadbeat dads)
• The fathers non-market capital is higher (men
  who enjoy childcare and are good at it new
  dads)
• The mothers market capital is higher (highly
  educated women in professional occupations)
• The mothers non-market capital is lower
  (career-oriented women, post-natal depression?)

11
Parental endowments and shared parenting

• Fathers who provide early years childcare are
• Less likely to have a degree or work in a skilled
  occupation
• More likely to have lower earnings capacity than
  their wives
• More likely to be under 30 at the birth of child
• More likely to live in the most deprived areas
  and in non-owner occupied housing (long hours
  only)
• More likely to be first-time dads
• More likely to smoke and have relatively negative
  attitudes to education and personal
  responsibility
• Do not differ from other fathers in physical or
  mental health
• More likely to attend antenatal classes and be
  excited by the prospect of fatherhood
• Mothers who share childcare are generally less
  differentiated than fathers. Post-natal
  depression does not appear to be a primary driver
  behind shared parenting
• There are substantial numbers of paternal carers
  in all household types, including the more
  affluent (e.g. fifth of long hours dads have
  degrees)

12
Child endowments and shared parenting

• Parental investment decisions are likely to
  respond to the innate characteristics of the
  child
• Family fixed effect (FFE) estimates difference
  out the common inherited component of siblings
  endowments, but impose strong data requirements
  (not fulfilled in ALSPAC) and may introduce
  severe bias
• Child fixed effect (CFE) estimates difference out
  any time-invariant child characteristics, but are
  only available if outcomes are observed before
  and after the treatment
• Our approach is to condition on measures of child
  health, temperament and developmental ability
  between birth and 6 months

13
Child endowments and shared parenting

• Temperament measures at 6 months

Temperament consists of relatively consistent,
basic dispositions inherent in the person that
underlie and modulate the expression of activity,
reactivity, emotionality and sociability. Major
elements of temperament are present early in
life, and those elements are likely to be
strongly influenced by biological factors.
Robert B. McCall, taken from Goldsmith et al
(1987), pp. 524

• Nine dimensions of temperament have been
  identified, including approach (response to new
  people/situations), adaptability (long-term
  adjustment to change), persistence (ability to
  persevere in a task) and activity (being still
  vs. moving about)
• Health measures at birth and immediately after
• Birth weight, gestation at birth, complications,
  general health in 1st 6 months
• Development measures at 6 months
• Gross motor, fine motor, communication and social
  skills at 6 months

14
Child endowments and shared parenting

• No significant differences in child gender or
  health at birth
• Stronger associations between child temperament
  in infancy and paternal childcare in Years 23
  than in Year 1. This suggests that parents
  decisions are affected by the temperament of the
  child, rather than the reverse
• Fathers appear slightly more likely to care for
  sons with easy rather than difficult
  dispositions, but we find fewer differences for
  daughters
• Sons who experience long hours of paternal care
  tend to score higher on the developmental
  measures, particularly the gross motor and fine
  motor sub-scores, and particularly if the care
  took place in infancy
• None of the differences were large. We conclude
  that child endowments do not appear to be an
  important determinant of shared parental
  childcare, although fathers may be slightly more
  likely to care for boys with relatively positive
  characteristics

15
Effects on school readiness and the role of
selection bias Unconditional OLS estimates

• All measures standardised to mean 100, standard
  deviation 10 on full sample available.
• Higher behaviour scores indicate greater
  behavioural problems
• Conditioning variables Childcare by
  family/friend and by paid carer Y1 Y23 (8
  dummies), childcare arrangements age 3 to school
  entry (6 dummies)

16
Entry Assessment Boys

• A. Childcare history B. Socio-economic
  endowments C Fathers personal endowments D.
  Mothers personal endowments E. Child endowments
• indicates rejection of null hypothesis H0 ?? 0
  
• , , and indicate significance at the 1, 5
  and 10 levels respectively

17
Entry Assessment Girls

• A. Childcare history B. Socio-economic
  endowments C Fathers personal endowments D.
  Mothers personal endowments E. Child endowments
• indicates rejection of null hypothesis H0 ?? 0
  
• , , and indicate significance at the 1, 5
  and 10 levels respectively

18
Behavioural problems Boys

• A. Childcare history B. Socio-economic
  endowments C Fathers personal endowments D.
  Mothers personal endowments E. Child endowments
• indicates rejection of null hypothesis H0 ?? 0
  
• , , and indicate significance at the 1, 5
  and 10 levels respectively

19
Behavioural problems Girls

• A. Childcare history B. Socio-economic
  endowments C Fathers personal endowments D.
  Mothers personal endowments E. Child endowments
• indicates rejection of null hypothesis H0 ?? 0
  
• , , and indicate significance at the 1, 5
  and 10 levels respectively

20
Interaction effectsLong hours of paternal care
in Years 23 on boys Entry Assessment

• Top coefficient is effect for group with
  characteristic, bottom coefficient for group
  without characteristic
• indicates rejection of null hypothesis that
  effects are equal
• , , and indicate significance at the 1, 5
  and 10 levels respectively

21
Interaction effectsMedium hours of paternal
care in Year 1 on boys Behavioural problems

• Top coefficient is effect for group with
  characteristic, bottom coefficient for group
  without characteristic
• indicates rejection of null hypothesis that
  effects are equal
• , , and indicate significance at the 1, 5
  and 10 levels respectively

22
Interaction effectsMedium hours of paternal
care in Year 1 on girls Behavioural problems

• Top coefficient is effect for group with
  characteristic, bottom coefficient for group
  without characteristic
• indicates rejection of null hypothesis that
  effects are equal
• , , and indicate significance at the 1, 5
  and 10 levels respectively

23
Interaction effectsMedium hours of paternal
care in Years 23 on boys Behavioural problems

• Top coefficient is effect for group with
  characteristic, bottom coefficient for group
  without characteristic
• indicates rejection of null hypothesis that
  effects are equal
• , , and indicate significance at the 1, 5
  and 10 levels respectively

24
Interaction effectsMedium hours of paternal
care in Years 23 on girls Behavioural problems

• Top coefficient is effect for group with
  characteristic, bottom coefficient for group
  without characteristic
• indicates rejection of null hypothesis that
  effects are equal
• , , and indicate significance at the 1, 5
  and 10 levels respectively

25
Some potential explanations

• Household income
• Parental relationship quality
• Conflict, e.g. shouting, hitting, throwing
  things, walking out of the house
• Mothers satisfaction, e.g. finances, time
  together, sex, major decisions
• Joint leisure activities, e.g. going out for a
  meal, visiting friends
• Communication, e.g. laughed together, kissed or
  hugged, discussed day
• Attitudes to parenthood
• E.g. Children are fun I enjoy watching my
  child develop I dislike the mess that
  surrounds my child I would have preferred not
  have had this child when we did
• Parenting behaviours and the home environment
• E.g. frequency to reading, singing, playing with
  child breastfeeding outings time with other
  children time watching TV toys and books
• Value-added specification
• Health, temperament and development between 15
  and 30 months

26
Entry Assessment Boys

• F. Parent and child endowments childcare
  history G. Household income H. Parental
  relationship quality I. Attitudes to parenthood,
  J. Parenting behaviours and the home environment
  K. Child outcomes 15-30 months
• indicates rejection of null hypothesis H0 ?? 0
  
• , , and indicate significance at the 1, 5
  and 10 levels respectively

27
Entry Assessment Girls

• F. Parent and child endowments childcare
  history G. Household income H. Parental
  relationship quality I. Attitudes to parenthood,
  J. Parenting behaviours and the home environment
  K. Child outcomes 15-30 months
• indicates rejection of null hypothesis H0 ?? 0
  
• , , and indicate significance at the 1, 5
  and 10 levels respectively

28
Behavioural problems Boys

• F. Parent and child endowments childcare
  history G. Household income H. Parental
  relationship quality I. Attitudes to parenthood,
  J. Parenting behaviours and the home environment
  K. Child outcomes 15-30 months
• indicates rejection of null hypothesis H0 ?? 0
  
• , , and indicate significance at the 1, 5
  and 10 levels respectively

29
Behavioural problems Girls

• F. Parent and child endowments childcare
  history G. Household income H. Parental
  relationship quality I. Attitudes to parenthood,
  J. Parenting behaviours and the home environment
  K. Child outcomes 15-30 months
• indicates rejection of null hypothesis H0 ?? 0
  
• , , and indicate significance at the 1, 5
  and 10 levels respectively

30
Conclusions Entry Assessment

• We find that boys who experience long hours of
  shared parental care when they are toddlers (20
  of the boys in our sample) perform more poorly on
  academic assessments on entry to school, and the
  magnitude of this effect is non-trivial
• We find no averse effects of shared childcare on
  girls Entry Assessment, and no adverse effects
  of shared childcare in infancy, or care of
  moderate intensity in the later period
• The negative effect we identify for boys cannot
  be explained by selection in the observable
  endowments of either parents or the boys
  themselves
• The effect appears robust across a wide variety
  of sub-groups of boys in different types of
  household. The exception is that it appears to be
  restricted to boys who did not experience any
  non-parental form of childcare when they were
  toddlers.
• The negative effect appears to emerge after the
  age of 2½ as we find that it is not reduced at
  all by controlling for a range of measures of
  health, temperament and developmental ability
  prior to this age
• Boys poorer academic ability does not appear to
  be accompanied by any increase in behavioural
  problems

31
Conclusions Entry Assessment

• Examination of a number of measures of the home
  environment do not provide any real explanation
  for this finding
• Attachment theory stresses the importance of
  mother-infant attachment, but the key period for
  attachments to form between 6 and 9 months, not
  in the toddler years
• Entry Assessment is comprised of teachers
  ratings of ability. It is possible that standards
  are higher in the schools attended by boys in
  this group, but given the diversity in family
  background of boys in this group this seems
  unlikely
• It is also possible that there is systematic
  teacher bias in assessments, but this would
  require that teachers observe some marker for
  shared parental childcare in the early years

32
Conclusions Behaviour

• Moderate hours of paternal childcare are
  associated with fewer behaviour problems in
  children, but only if that care begins after the
  first year of life (25 of cases).
• Moderate hours that begin in Year 1 but then stop
  (only 4 of cases) are associated with small
  adverse effects on behaviour, particularly in the
  sons of mothers who are not employed, suffer PND
  and who have low quality mother-child
  interactions.
• Moderate paternal care that is maintained over
  the 3 years (11) is not associated with any
  differences compared to maternal-only parental
  care.
• We find similar effects of boys and girls,
  although the effects for girls are smaller and
  less precisely estimated
• Long hours of paternal care, either of infants or
  of toddlers, are not associated with any effect
  on behaviour compared with maternal-only parental
  care.
• These effects appear to have already emerged to
  some degree by age 2½. It is possible that these
  results reflect selection in terms of child
  endowments that is not apparent at 6 months.

33
EXTRA SLIDES
34
Is there heterogeneity in the effects of shared
parental care? Other results
35
(No Transcript)