CHILD POVERTY IN THE UK: SOCIO-DEMOGRAPHIC SCENARIOS TO 2020 FOR CHILDREN

1
CHILD POVERTY IN THE UKSOCIO-DEMOGRAPHIC
SCENARIOS TO 2020 FOR CHILDREN

• Phil Rees John Parsons
• University of Leeds
• Paper presented at the
• Third International Population Geographies
  Conference
• University of Liverpool, Liverpool, UK
• 19-21 June 2006

2
Context (1) Questions asked

• The work was part of a School of Geography, Leeds
  (SOG) project for the Joseph Rowntree Foundation
  (JRF) in collaboration with the Institute for
  Fiscal Studies, London (IFS)
• JRF wished to know answers to these questions
• Will the government hit its child poverty
  reduction targets?
• Are socio-demographic trends favourable for
  achieving these goals or not?
• How will poverty alleviation strategies affect
  children in different parts of the United
  Kingdom?

3

• Child poverty
• living on less than 60 of UK average (median)
  income (household income standardised for number
  of adults children)
• AHC After Housing Costs

Tony Blair (1999) this government will
eradicate child poverty by 2020 IFS interprets
government statements to mean 1998/9 to 2004/5
25 reduction in numbers 1998/9 to 2010 50
reduction in numbers 2010 to 2020 50
reduction in numbers (not official)
These are really ambitious targets Will
socio-demographic trends help?
4
Context (2) What was needed?

• IFS method is to run a micro-simulation model for
  households, families and children using inputs
  from the latest Family Resources Survey (DWP/ONS)
• IFS needed to re-weight the micro-population
  using key population variables projected to 2010
  and 2020 and SOG
• IFS/SOG might have created a dynamic household
  microsimulation model but time and resource
  limits meant this was not possible (though two
  teams in SOG are doing this in current projects
  with EPSRC and ESRC support)

5
Our pragmatic approach

• Use the Individual Sample of Anonymised Records
  from the 2001 Census of Population to create an
  8-dimensional population array P2001(x1, , x8)
• Develop 8 sub-array projections for 2010 and
  2020 P2010(x1,x2), P2010(x3) etc marginal
  distributions of the full array
• Apply Iterative Proportional Fitting to adjust
  the 2001 population array to the 2010 and 2020
  constraints
• Because the data came from many different sources
  and despite adjusting each sub-array to the same
  grand UK population total, we were not satisfied
  with the robustness of the results (further work
  planned to resolve this by Parsons, Jin and Rees)
• Nevertheless, the marginal distributions were
  considered reasonable projections of each
  dimension for IFS to use them to re-weight their
  microsimulation results
• Full reports will be published by the Joseph
  Rowntree Foundation this summer

6
The seven population dimensions projected for
each region

• Age and sex, with child dependency
• Ages 16-18 split between dependent and
  non-dependent children
• Household size
• Large households are at higher risk of poverty
• Number of dependent children
• Families with many children tend to be poor
• Family type
• Some types e.g. lone parent families are at high
  risk of poverty
• Ethnic groups
• Some groups face higher poverty risk than others
• Number of earners in household
• Non-earner households are poorer
• Tenure
• Households in some tenure types (e.g. social
  housing) tend to be poor

7
Methods

• A mix of methods was used to project these
  marginal dimensions into the future
• Mostly we used extrapolation based on trends
  1981-2001 in census distributions
• Sometimes we used trends in FRS distributions for
  1998-2004
• For ethnic groups we built a cohort-component
  demographic projection model
• The paper goes into the detail of data sources
  and methods
• In the rest of the presentation, attention is
  focussed on our results and their implications
  for child poverty

8
The 2004-based UK projections (the latest) see
population growth continuing until the 2070s,
despite below replacement fertility, because of
higher migration assumptions and improving
longevity. But the population continues to age
and the number of children fluctuate (echoes of
echoes of the baby boom).
The number of dependent children is falling in
this decade but not in the next (in part because
of a small shift from non-dependency to
dependency among 16-18 year olds). The numbers
help attainment of the 2010 target but not the
2020.
9
These figures combine GAD 2004 national
projections with 2003 regional projections for
England. Southern regions are projected, because
of lower mortality, higher immigration and net
internal migration gains in younger ages, to grow
faster than Northern. Note that the populations
of the North East and of Scotland grow hardly at
all.
10
These projected population changes shift the
distribution of children towards regions with
higher incomes (after housing costs). This will
be helpful for the achievement of the child
poverty reduction goals.
11
Our projections of population by household size
(based mainly on extrapolating 1981-2001 trends
because the national projections fail to provide
much information!) show a continuing fall in the
numbers of larger households (with 4, 5 or 6
persons). Again, this trend favours attainment of
child poverty goals.
12
We use a set of cartograms to display the changes
in the population variables. The cartogram
assigns an area on the map in proportion to the
regional population (developed by Durham, Dorling
and Rees). The cartogram gives due population
weight to the urban region of London.
13
These maps show a uniform and substantial fall
(25) in the percent of people living in larger
households between 2001 and 2020.
14
The graph and maps confirm that the fall in
household sizes is due to both fewer people
living in households with dependent children and
a fall in share of households with 3 or more
dependent children
15
Lone parent families are the poorest and our
projections show a reduction in the share of the
population living in these households, except in
London.
16
The White population grows a little during the
2001-2020 period (mainly because of new
immigration). The Ethnic Minority population
grows very substantially because of demographic
momentum and high immigration. However, there
are differences between the groups in terms of
(child) dependents. Very little growth of
dependent children in the Black group, high
growth in Asian and Mixed dependent children.
17
This graph for the Asian population shows the
growth in dependent children but note that there
is higher relative growth in the labour force and
older ages (as a result of the ageing of earlier
cohorts). Fertility rates in the Asian groups are
converging to lower national levels.
18
Our projections show what changes are likely in
the ethnic composition of the regions of the
country. There are profound contrasts in ethnic
diversity between the South East regions and
the Celtic periphery. The changing population mix
will tend to work against the attainment of child
poverty reduction goals as the population shifts
towards groups with larger and poorer populations
on average.
19
What is interesting about this map is that the
greatest relative change in regions that have
lower concentrations. There is deconcentration of
the ethnic minority population at regional scale
(and also probably within regions and cities
see papers by Simpson). The implications for
child poverty reduction are indirect. The ethnic
minority populations growing fastest outside
their areas of concentration are probably better
off and suffer less from child poverty, so this
is a favourable trend.
20
No great changes projected here, but this ignores
other influences on the number of earners such as
rising age at retirement, continuing rise in
female labour force participation and measures to
encourage lone parents and working age persons on
disability benefit into work.
21
It is difficult to be confident about the
flatness of the trends here. This is based on
extrapolation of two decades of Conservative and
Labour government. Imagine the contribution to
national wealth and child poverty reduction if
all regions could attain the labour force
participation of the East of England (Cambridge,
Peterborough, Norwich, Ipswich etc).
22
Our projections see continuing reduction in the
social rent household category. This is helpful
for reaching child poverty reduction targets.
However, London appears to be an exception, with
a rising proportion in social rented housing.
23
Conclusions

• The governments child poverty reduction target
  is tough and it has missed it 2004/5 target
  (though not by much).
• The reductions will need to accelerate to achieve
  2010 and 2020 targets
• Socio-demographic trends are favourable in small
  ways (because the targets are defined in number
  rather than percent terms).
• However, some trends move in the opposite
  directions so the net effect is unclear. We need
  to be confident about our SAR/IPF model or in
  future dynamic micro-simulation models to be
  certain what the overall outcome will be.
• When the IFS work is published, we will be able
  to judge the effect of socio-demographic trends
  against anticipated or suggested policy changes.