EVIDENCE ON NEIGHBOURHOOD EFFECTS
AND NEIGHBOURHOOD POLICY IN THE UK:
ARE THEY CONNECTED AND SHOULD
THEY BE?
ESRC seminar "Neighbourhood effects, neighbourhood based
problems and international policy solutions”, 7-8th April 2011,
Urban Studies Department, University of Glasgow
Rebecca Tunstall, LSE
1
Introduction
A.
Some new evidence on neighbourhood effects
- Tunstall, R; Lupton, R; Kneale, D and Jenkins, A (2011a) Teenage
housing and neighbourhoods and adult outcomes: Evidence from the
1970 cohort study CaseReport No.64 London: CASE
- Tunstall, R; Lupton, R; Kneale, D and Jenkins, A (2011b) Growing
up in social housing in the new millennium: Housing,
neighbourhoods and early outcomes for children born in 2000
CasePaper No.143 London: CASE
- Postcode discrimination in employment - forthcoming study
B. Discussion of the relationship between evidence of
neighbourhood effects and policy
C. An experiment!
2
A. Introduction
The first pair of studies develop the findings of two previous
reports (Feinstein et al, 2008, Lupton et al, 2009).
- The earlier reports examined housing in childhood for those born
in 1946, 1958, 1970 and 2000, and its relationship with adult
outcomes across a range of measures (except for those born in
2000).
- They found as yet unexplained connections between being ‘ever’
in social housing in childhood and worse adult outcomes on an
overall measure of deprivation and a range of individual measures
for those born in 1958 and in 1970 (but not for those born in 1946)
(Feinstein et al, 2008, Lupton et al, 2009).
3
The aims of the research
The pair of studies uses:
1) the British Cohort Study, whose members were born in 1970 and
2) the Millenium Cohort Study whose members were born in 2000.
One of the aims of the pair of projects was to explore whether
features of social housing neighbourhoods might constitute either
all or part of what appeared to be a ‘tenure effect’.
The studies were funded by the Homes and Communities Agency
and the Tenant Services Authority, and completed with help from
colleagues at the Centre for Longitudinal Studies at the Institute of
Education, who assisted in linking cohort data to neighbourhoods.
4
1st study: Using the British Cohort Study
The British Cohort Study follows a sample of people born in Great
Britain in 1970 as they have grown up. Data were collected at birth,
at age 5, and 10, 16, 26, 30, and 34.
Childhood waves of data collection involved interviews with parents.
Children were also tested and as they entered adolescence they
were interviewed.
5
The British Cohort Study
We report data for a sub-sample of cohort members:
•Born in England, rather than Great Britain (in contrast to Feinstein
et al, 2008 and Lupton et al, 2009);
•Included in data collection at three points, in 1970, 1986 and 2004;
•For whom full data was collected on the main variables used in
this analysis.
This produces a sample of about 4,000 people in all (the British
Cohort Study included about 9,000 people born in 1970 who were
tracked for at least some variables all the way through to 2004,
across the whole of Great Britain).
6
Characterising teenagers’ neighbourhoods
We used the 1991 Census to characterise cohort members’ 1986
neighbourhoods. (This was the closest date match available;
other research suggests that the deprivation of neighbourhoods
relative to others changes slowly).
We looked at:
• Small neighbourhood scale (Enumeration Districts, since
replaced by Output Areas, about 100-150 households); and
• Large neighbourhood scale (electoral wards, about 2,200
households).
We examined:
• Neighbourhood deprivation: what decile of neighbourhoods the
cohort member’s address was in using the ‘Carstairs’ index of
deprivation (similar to the IMD, but available for 1991); and
• Neighbourhood tenure mix: whether or not the neighbourhood
was ‘social housing dominated’
7
Neighbourhood deprivation: Proportion of teenagers in each
tenure found in small neighbourhoods in different deciles of our
deprivation index 1991, at age 16 in 1986
30
Percentage in this kind of neighbourhood
25
20
15
10
5
0
Most
deprived
10%
2nd most
3rd
4th
Social renting
5th
6th
Private rented and other
7th
8th
Ow ner occupied
9th
Least
deprived
10%
8
Characterising outcomes in adulthood
We examined outcomes for cohort members at the age of 34 in
2004.
We used 11 measures of outcomes:
1)
In paid employment/not;
2)
Claiming means tested benefits/not;
3)
Literacy or numeracy problems/none;
4)
Depressed/not;
5)
Suffering malaise (mild depression)/not;
6)
Taking regular exercise/not;
7)
Low life satisfaction/not;
8)
Smoking at least 1 cigarette a day/not;
9)
Highest qualifications;
10) Self-rated health; and
11) Self-efficacy.
9
Odds ratios for raw outcomes at age 34 for those in both rented
tenures as teenagers, compared to those in home ownership
Exercise
Health (Improvement)
Employed
Highest Qualification (Additional)
Low Life Satisfaction
Malaised
Literacy or Numeracy Problems
Low Self Efficacy
Depressed
Smokes cigarettes
Received Means Tested Benefits
0
1
2
3
4
5
6
Odds Ratio
Social Housing
Private Rental
10
The role of teenage housing tenure in
explaining adult outcomes: Controls
We controlled for individual and family advantage, with 2 sets of
controls.
The small set of controls included 8 measures:
1)
Index of Family Advantage;
2)
Mothers’ age at the birth of her first child;
3)
Whether mother had moved away from her home region by
the birth of first child;
4)
Cohort member’s weight at birth;
5)
Whether cohort member was born into a lone parent family;
6)
Whether cohort member was in a lone parent family at 16;
7)
Number of siblings resident in the household at 16; and
8)
Region lived in at 16.
The large set of controls included 56 variables, as used in Feinstein
et al (2008) and Lupton et al (2009).
11
Odds ratios for outcomes at age 34 for those in social
rented housing as teenagers, compared to those in
home ownership as teenagers (raw outcomes and
various controls)
Exercise
)Health (Improvement
Employed
)Highest Qualification (Additional
Low Life Satisfaction
Malaised
Literacy or Numeracy Problems
Low Self Efficacy
Depressed
Smokes cigarettes
Received Means Tested Benefits
0
0.5
1
1.5
2
2.5
3
3.5
4
12
Odds Ratio
No Controls
Index of Advantage
Small set of controls
4.5
Large set of controls
The role of teenage neighbourhoods in
explaining adult outcomes…
13
Exercise
Health
(Improvement)
Employed
Highest
Qualification
(Additional)
Low Life
Satisfaction
Malaised
Literacy or
Numeracy
Problems
Low Self Efficacy
Depressed
Smokes cigarettes
Received Means
Tested Benefits
0
0.5
1
1.5
No Controls
Small set and larger area deprivation
Small set and larger area tenure mix
2
2.5
3
3.5
Small set
Odds
Ratio
Small set and smaller area deprivation
Small set and smaller area tenure mix
4
4.5
14
After applying controls for family and individual circumstances with
both the small and large sets of controls, taking neighbourhood
characteristics into account modified somewhat* the relationship
seen between teenage tenure and adult outcomes.
Thus, the associations between tenure and outcomes we have
found can be described as likely to be partly* due to the
characteristics of the neighbourhoods that homes in different
tenures were found in:
–in terms of deprivation; and
–in terms of tenure mix;
–at both small neighbourhood scale (about 100-150 households)
and larger neighbourhood scale (about 2,200 households).
15
2nd study: Using the Millennium Cohort
Study
The Millennium Cohort Study is a longitudinal study of children
born in 2000. Almost 15,000 children have been tracked to the age
of 5 in 2006.
The 2006 study includes interviews with parents and tests on
children.
We report data for cohort members born in England, rather than
Great Britain (in contrast to Feinstein et al, 2008 and Lupton et al,
2009).
16
Tenure and neighbourhoods at age 5 in 2006
We used the Index of Multiple Deprivation 2007 to characterise the
Lower Super Output Areas (LSOAs, neighbourhoods typically
with 600 homes and 1,500 people) in which cohort members
were living in at age 5.
Again the situation of children in social housing stands out.
17
Percentage of children in each housing tenure in LSOAs in
Proportion of 5yr olds in each tenure in this kind of nhood
different deciles of the IMD 2007, at age 5 in 2006
30
25
20
15
10
5
0
Most
2nd most
deprived
10%
3rd
4th
5th
Social renting
6th
Private rented
7th
Owned
8th
9th
Least
deprived
10%
18
Early outcomes at age 5 in 2006
We examined early developmental outcomes through scores on
tests of vocabulary and drawing skills at age 5.
19
0
20
40
60
Mean test scores by housing tenure, age
5 in 2006
owned
LA/HA rent
naming vocabulary
private rent
pattern construction
weighted estimates
20
0
20
40
60
Mean test scores by neighbourhood IMD
2007 decile, age 5 in 2006
poorest
2
3
4
5
naming vocabulary
weighted estimates
6
7
8
9
richest
pattern construction
21
The role of housing tenure and
neighbourhoods in early outcomes
We carried out regression analysis on test scores to control for a
small number of factors:
i.
Whether or not children lived in the most deprived tenth of
neighbourhoods;
ii.
Index of Family Advantage )based on parents’ education and
occupational status);
iii.
Mother’s age at first birth;
iv.
Family structure;
v.
Number of siblings.
22
Regression results
Half* the gap in test scores between children in both rented tenures
and those in home ownership was removed using the small
set of controls
Thus, the typically high deprivation levels of social housing
neighbourhoods, included in these controls, appear to
constitute part* of the link seen between tenure and early
childhood outcomes.
Individual and family factors also appear to constitute part* of the
link.
23
Summary of both studies
The typically high deprivation levels of social housing
neighbourhoods (and for the 1st study, concentrations of social
housing) appear to constitute part* of the link seen between
social rented tenure and various outcomes for two generations.
24
3rd study: Experimental correspondence test
for postcode discrimination in employment
Discrimination by outsiders including in employment one of
main hypothesised mechanisms for neighbourhood effects
Residents of deprived areas, their advisors, and members of the
public believe this discrimination takes place - and some
employers have admitted to it…
Results due late 2011…
25
B. The relationship between evidence of
neighbourhood effects and neighbourhood
policy
Has 1st, 2nd project evidence been connected to policy (to date)?
No.
No ‘policy implications’ in reports
UK housing and neighbourhoods policy environment 2009-2010
not ready for research
Should this research be connected to policy?
Perhaps not.
26
A very basic model of policy making and the
role of evidence in it…
1)
P is a social/economic problem
2)
Is P associated with/possibly caused by factor X?
(a role here for specialist factor X researchers)
3)
If yes, )specialist researchers’) policy implication =
change/remove X
27
HM Treasury’s model of policy making and
the role of evidence…
HM Treasury’s Green Book constitutes binding guidance for all
central government departments and agencies in the UK:
“all new policies, programmes and projects, whether revenue,
capital or regulatory, should be subject to comprehensive but
proportionate assessment…
The essential technique is option appraisal, whereby
government intervention is validated, objectives are set, and
options are created and reviewed, by analysing their costs
and benefits” )HMT 2003 p1-3).
28
Have neighbourhood policies been based on
evidence about neighborhood effects to
date?
Eg UK 1997-2010:
Largely, no:
- Absence of timely evidence
- Presence of other rationales
29
Have neighbourhood policies been likely to
have any effect on neighbourhood effects?
Eg. UK 1997-2010:
Some, if limited, effects on deprived neighbourhood outcomes in
education, worklessness…
Some, limited, change in characteristics of neighbourhoods we
attribute neighbourhood effects to:
-population composition
-neighbourhood infrastructure including services
-relationship with other areas.
Not clear what critical thresholds are and if they have been passed,
or how many neighbourhoods
Many policies compensated for individual deficits
30
Housing, neighbourhood policy implications
of evidence on neighbourhood effects:
Is it any of our business?
Neighbourhood effects can be effects on anything…
How much should housing and neighbourhood policy attempt to
alter outcomes such as: paid employment, means tested
benefits, literacy or numeracy problems, depression, exercise,
life satisfaction, smoking, qualifications, self-rated health, selfefficacy (1st study); pattern construction tests at age 5 (2nd
study)?
Can it be our business alone - do we understand enough about
these outcomes?
31
Is it worth trying?
Neighbourhood effects are ‘small’
How much social resource should we devote to trying to change test
scores at age 5 (2nd study)?
- Results at age 5 on similar tests were associated with*
achievement in reading and maths at age 10 and qualifications and
wages at age 30 (Feinstein and Duckworth, 2006).
- Children in most deprived decile of neighbourhoods scored about
1.5 points lower* than other after controls.
These are questions for policymakers – but they need help.
32
Is something else
also/more worth trying?
Housing and neighbourhood policies – especially the highest
economic/social cost ones - are not the only options
•Eg remedial not preventative policies – for 1st study outcomes,
employment support, smoking cessation, literacy classes, exercise
clubs…
•Eg for 2nd study outcomes...
…child test scores are also associated with gender (Schoon et al, 2010),
ethnicity (Dearden and Sibieta, 2010), parents’ employment status and
financial situation, parenting practices, parents’ relationship quality, motherchild relationships and mother’s wellbeing and self-esteem (Jones, 2010)
33
Lo
ne
pa
ep
ta
y
3
ild
(c
o
0+
re
d
to
1
30
si
)
b)
-3
4
)
)
nt
s
ni
ng
pa
re
ow
to
2
m
p
to
m
pa
(c
o
m
p
to
)
9)
-3
4
er
30
ot
h
)
1)
ni
ng
to
m
os
t)
ow
to
to
p
m
p
om
(c
o
(c
m
p
to
m
p
to
(c
o
m
p
m
p
(c
o
(c
o
ed
he
r4
ge
re
nt
e
20
cil
he
ru
ch
M
ot
ta
O
nl
re
n
at
e
M
ot
de
(c
o
s
(c
o
sib
tile
3+
in
i
te
d
qu
lr
en
ed
riv
Pr
iv
td
cia
fa
m
ily
So
ta
dv
as
M
os
Le
Test score gap
Gaps in pattern construction test scores for Millenium Cohort children in
different situations, after controls for neighbourhood deprivation, family
advantage and other individual and family factors, age 5 in 2006
5
4
3
2
1
0
34
Finally: An emerging gap between
researchers and policymakers?
“the conventional policy wisdom is… so dominant” )Galster 2007
p524),
But
the ‘policy thrust’ towards creating more socially mixed
neighbourhoods for poorer residents “has not been without its
skeptics” (2007 p524)
… Kearns, Atkinson, Kearns, Meen, Musterd, Ostendorf, Freidrichs,
Cole, Goodchild, Delorenzi…
35
And subsequently:
“the main sources of low incomes are to be found in earnings,
employment, and demographics, not in neighbourhood
characteristics” (Bolster et al. 2007 p234)
“neighbourhood effects are small compared to the effects of
individual and family characteristics" (van Hamm and Manley 2009
p)
“evidence…. does not support the conclusion that NEs are
quantitatively all that important… the conclusion for policy is to
reduce overall income inequality rather than to promote mixed
communities” (Cheshire 2007 p34)
36
Bond et al’s comments
“It seems likely… that the reviews [of evidence on mixed
tenure] used by policymakers could be less thorough or
rigorous than they might be” (p70).
Bond, L; Sautkina, E and Kearns, A )2010) ‘Mixed messages
about mixed tenure: Do reviews tell the real story?’ Housing
Studies 26(1) pp69-91
37
The UK Social Research Association’s view
of policy making and the role of evidence…
“Social researchers may not be in a position to prevent action
based on their findings. They should, however, attempt to
pre-empt likely misinterpretations and should counteract them
when they occur” (SRA 2003 p18).
38
C. Action research!
Q. Is there a latent consensus about the implications of evidence
on neighbourhood effects for housing and neighbourhoods
policy, and if so, what is it?
A ‘Delphi’ review )Linstone and Turoff 1972)
1) Gathering individual views;
2) Feedback to individuals of collated, anonymised individual
views;
3) Feedback from individuals on collated views…
39
Is future neighbourhood policy likely to have
an effect on neighbourhood effects?
What is a ‘neighbourhood policy’? The most important
neighbourhood policies are mainstream public policies, many
incorporating some redistribution and/or or anti-segregation effects
(and private policies)
UK’s new neighbourhood policy 2010-:
Another experiment
- Withdrawal of most ABIs
- Spending cuts
- ‘Localism’
Potential change in characteristics of neighbourhoods we attribute
neighbourhood effects to:
-population composition
-neighbourhood infrastructure including services
-relationship with other areas
40
References
Dearden, L and Sibieta, L (2010) 'Ethnic Inequalities in Child Outcomes.' In K. Hansen, H. Joshi and S. Dex,
Children of the 21st Century: The first 5 years, Bristol: Policy Press, pp 169-184.Jones, 2010
Feinstein, L. and Duckworth, K. (2006) Development in the early years: Its importance for school performance
and adult outcomes Wider benefits of Learning, Research Report no. 20, London: Centre for Research on the
Wider Benefits of Learning.
Feinstein, l., Lupton, R., Hammond, C., Mujtaba, T. and Sorhaindo, A. with Tunstall, R., Richards, M., Kuh, D.
and Jonson, J. (2008) The public value of social housing: A longitudinal analysis of the relationship of housing
and life chances, London: Centre for Research on the Wider Benefits of Learning, Institute of Education,
University of London.
Galster, G )2007) ‘Should policymakers strive for neighbourhood social mix? An analysis of the Western
European evidence base’ Housing Studies 22(4) pp523-545
HM Treasury (2003) The Green Book: Appraisal and evaluation in central government: Treasury Guidance
London: TSO
Linstone, HA and Turoff,M (2002) The Delphi method: Techniques and applications (downloaded from
http://is.njit.edu/pubs/delphibook/ch1.html, orginally published 1972)
Lupton, R.; Tunstall, R.; Sigle-Rushton, W.; Obolenskaya, P.; Sabates, R.; Meschi, E.; Kneale, D. and Salter, M.
(2009) Growing up in social housing in Great Britain: The experience of four generations, London: Homes and
Communities Agency.
Mullins, D (2006) ‘Exploring Change in the Housing Association Sector using the Delphi Method’ Housing
Studies, 21 (2), 227-251
Schoon, I., Cheng, H. and Jones, E. )2010) ‘Resilience in children’s development’ in K. Hansen, H. Joshi and S.
Dex (eds) Children of the 21st century: The first 5 years, Bristol: Policy Press, pp 235-248.Jones, E. (2010)
‘Parental relationships and parenting’, in K. Hansen, H. Joshi and S. Dex, )eds) Children of the 21st century: The
first 5 years, Bristol: Policy Press, pp 53-75.
Social Research Association (2003) Ethical guidelines (downloaded March 2011 from http://www.thesra.org.uk/documents/pdfs/ethics03.pdf)
Tunstall, R; Lupton, R; Kneale, D and Jenkins, A (2011a) Teenage housing and neighbourhoods and adult
outcomes: Evidence from the 1970 cohort study CasePaper No.64 London: CASE
(http://sticerd.lse.ac.uk/dps/case/cr/CASEreport64.pdf)
Tunstall, R; Lupton, R; Kneale, D and Jenkins, A (2011b) Growing up in social housing in the new millennium:
Housing, neighbourhoods and early outcomes for children born in 2000 CasePaper No.143). London: CASE
(http://sticerd.lse.ac.uk/dps/case/cp/CASEpaper143Tunstall.pdf)
41