Neighbourhood Effects and the Welfare State. Towards a European research agenda? Roger Andersson Institute for Housing and Urban Research Uppsala university, Sweden

1
Neighbourhood Effects and the Welfare State.
Towards a European research agenda?Roger
AnderssonInstitute for Housing and Urban
ResearchUppsala university, Sweden

• Paper for the conference Neighbourhood Effects
  Studies on the Basis of European Micro-data
• at Humboldt University of Berlin on March 29 and
  30, 2007

2
Contents

• Firstly, I introduce my current research
  programme, a programme I believe also guides many
  colleagues engaged in studying residential
  segregation and the issue of neighbourhood
  effects (NE).
• Secondly, I will deal with three particular
  aspects of NE that need further and elaborated
  studies.
• Thirdly, as I know this is a conference focusing
  on micro data, my presentation also contains
  information about the Swedish statistical
  registers and how such data can be used for
  extending our knowledge about segregation
  dynamics and effects of segregation on the social
  trajectories of individuals.

3
A research programme on neighbourhood mix and
neighbourhood effects
Global, National and Urban Contexts
The Micro Structure of the Housing Stock
(neighbourhoods composition in terms of tenure
and housing types)
(1)
Social and Ethnic composition of neighbourhoods
(2)
(3)
Social interaction Effects on attitudes and
behaviour
Social opportunities
(4)
4
Summary of questions

• Is there really a strong relation between housing
  mix and social mix? This is a fundamental issue
  since planning for social mix is based on the
  assumption that the micro structures of the
  housing stock in terms of tenure, housing types,
  size and cost of dwellings etc are thought to
  strongly influence the population composition of
  neighbourhoods. (Relation 1)
• How does population composition of neighbourhoods
  affects residents social interaction and
  behaviour? (Relation 2)
• Are social opportunities related to peoples
  neighbourhood context? (Relation 3)
• If there is such a relation, to what extent is
  this produced/mediated by local social
  interaction? (Relation 4) The idea is that social
  opportunities might be directly or indirectly
  affected by residency.

5
Three equally important questions arise if one
wants to study these relations

• What population mix matters?
• What scale matters?
• What time matters?

6
Data sourcesSwedish social scientists,
especially segregation researchers, have access
to internationally unique types of data. I will
briefly describe the basic features of these
data. Four characteristics are of key importance

• 1. A personal ID code (personnummer) is used in
  all official registers. A similar code is used
  for firms. The individual-specific ID code
  comprises 10 digits and is given to everyone upon
  birth or immigration (permanent residents). This
  code is used by Statistics Sweden in all
  individual registers, such as the employment,
  income, population, education, and the event
  registers (birth, death, immigration,
  emigration).
• 2. There are constantly updated address registers
  (Register över totalbefolkningen, RTB) linked to
  the ID code mentioned in (1).
• 3. A geo-coded real estate and property register
  exists, linked to the address register
  (fastighetsregistren). The geo-coding of all real
  estates took several decades to finish, and this
  crucial part of the registers was not completed
  until about 1990.
• 4. The law grants researchers a reasonable easy
  and inexpensive access to data on individuals.

7

• By merging (1), (2), and (3) all residents in
  Sweden can be localised both in terms of housing
  and work places. This allows for the study not
  only of static distributions at any point in time
  but also of longer-term developments. An
  individuals housing and employment career can
  thus be studied both in its social and geographic
  context. Obviously, both migration and commuting
  can be studied using complete populations. If a
  person moves, this will show up in the address
  register and due to the fact that all addresses
  refer to specific and geo-coded buildings, the
  exact origin and destination location will be
  known.
• One obvious advantage is that data can be
  generated for researcher-defined geographical
  areas, escaping the sometimes not so relevant
  formal geographical, administrative divisions.

8

• It is not difficult to realise that these data
  are sensitive, and the use is restricted in
  several ways. However, there is an important
  paragraph in the Swedish data security
  legislation saying that access to the registers
  should be generously provided to researchers.
• Applications from researchers are scrutinised by
  a special committee at Statistics Sweden, and
  also by regional research ethics committees, who
  decide whether permission should be given and if
  certain restrictions should apply. Some
  restrictions are of a more general character, for
  instance that data on individuals or firms
  provided to researchers never contain the
  explicit ID code and that specific individuals
  should not be identifiable in publications.
• Furthermore, the most detailed geocodes
  (coordinates) are seldom provided, and
  researchers normally have to settle with 100m by
  100m coordinates (which of course is still a very
  detailed level). There are often also
  restrictions on handing out specific codes for
  the country of birth information.

9
Back to the research questionsHousing mix and
social mix

• In short, the idea is that housing mix (a mix of
  housing types and tenure types) will create
  social mix (a mix of households according to
  their socio-economic position) and that this will
  create better social opportunities for
  individuals. In fact, this standpoint is based on
  two crucial assumptions. The first is that social
  mix really enhances the individual opportunities
  (i.e. relations 3 and/or 4 in figure 1 are true).
  The second is that there is a strong relation
  between social mix and housing mix (relation 1 is
  true).
• Although these assumptions may be
  realistic/plausible, they are empirical questions
  that social science research needs to address.

10
The recently out-voted Swedish Social Democratic
government kept on reiterating the importance of
housing and social mix, echoing a position
established in the mid 1970s mixed tenure in
each and every neighborhood in our country is
something I aim for. No matter if I discuss from
the perspective of social (class) aspects or
(ethnic) integration aspects, it is of vital
importance that there is other than rental tenure
in our large (suburban) housing estates and that
the inner cities comprise not only private
ownership of apartments but also rental housing.
(Mona Sahlin, Minister of Housing and the Built
Environment, talking in the Swedish Parliament,
November 15, 2004
11

• The minister continued, now talking about areas
  having a concentration of low income households
  and a high proportion of unemployed people
• It should be good housing in safe and well
  serviced neighbourhoods like we find in other
  parts of the cities. These areas should be
  characterized by great variety, which implies
  that it should be possible but also interesting
  for households having an above average income
  level to live there. (Mona Sahlin, Minister of
  Housing and the Built Environment, talking in the
  Swedish Parliament, November 15, 2004)

12
Policy program theory?

• The minister does not explicitly say why this is
  important but other documents support the
  conclusion that Swedish ideas about mixing are
  based (albeit not entirely) on the assumption
  that there are social (negative and positive)
  externalities of different types of population
  mix. (See also Galster 2007)

13
Housing mix and social mix Evidence

• Musterd Andersson (2005) find that relation (1)
  (see figure 1) is rather weak in Sweden as a
  whole. Further study is needed, not least studies
  that analyse the relation more in detail for
  cities of different size. One may hypothesize
  that although the relation is quite weak at the
  national level it might very well be much
  stronger in the larger cities.
• Most socially homogenous neighbourhoods are
  middle class, home ownership areas. The rich are
  spatially more concentrated than are the poor.
  The debate on mix disregards this fact and
  focuses on concentrated poverty and immigrant
  dense areas and how more mix could be achieved in
  such areas.

14
Social and ethnic mix and neighbourhood effects

• Many researchers make use of Charles Manskis
  (2000) distinction between three types of
  neighbourhood effects endogenous, contextual
  (exogenous) and correlated. (See Galster, 2006).
  If we face endogenous interactions, the
  propensity of an agent to behave in some way
  varies with the behaviour of the group. In
  contextual interactions, the propensity of an
  agent to behave in some way varies with exogenous
  characteristics of the group members. Correlated
  effects concern situations when agents in the
  same group tend to behave similarly because they
  have similar individual characteristics or face
  similar institutional environments.

15
Endogenous exogenous effects relation 2 and 4
correlated effects relation 3 plus the wider
context.
Global, National and Urban Contexts
The Micro Structure of the Housing Stock
(neighbourhoods composition in terms of tenure
and housing types)
(1)
Social and Ethnic composition of neighbourhoods
(2)
(3)
Social interaction Effects on attitudes and
behaviour
Social opportunities
(4)
16
Correlated effects and the Welfare State a
proposition

• In countries where resources are fairly well
  distributed (reallocated over the tax system)
  differences in economic standard between
  households are less pronounced. Less social
  polarisation normally means less socio-economic
  segregation and also less spatial differences
  with regard to institutional quality.
• In cities with a well developed and reasonably
  priced local transportation system, labour market
  spatial mismatch can be expected to be of
  relatively little importance for the prospects of
  gaining and keeping employment.

17
What population mix matters?

• Might of course depend on what type of outcome we
  are studying.
• For individual income development we employed a
  multivariate model on Swedish data to estimate
  the relative importance of 4 types of mix
  dimensions (income, education, ethnicity, housing
  characteristics) and for each we tested different
  operationalisations (such as the relative size of
  advantaged and disadvantaged groups, entropy
  values, ratios). We found the importance of the
  neighbourhood income structure to be bigger than
  the other three. Effects are bigger for males
  than for females and for metropolitan areas
  compared to non-metropolitan areas.
• Neighbourhood definition SAMS (average pop.
  Size 1000).
• Sample size not a sample, but the entire adult
  Swedish population. Longitudinal 1991-1999
  material.
• Control for a range of time-invariant and
  variant personal and household demographic and
  socioeconomic characteristics, and Labour Market
  characteristics.
• Andersson, Musterd, Galster, Kauppinen (fc)

18
What population mix matters? (2)

• We ran similar models for exploring the
  importance of ethnic clusters in metropolitan
  Sweden for the income development of individuals
  in seven larger immigrant groups. We found the
  effects of own group concentrations to be
  negatively correlated with earning prospects if
  unemployment levels in ethnic clusters were
  modest or high.
• Neighbourhoods bespoke, 500m grid areas, 250m
  around each individual.
• Sample size not a sample but all adults residing
  in metropolitan Sweden 1995-2002.
• Musterd, Andersson, Galster, Kauppinen (fc)

19
Municipal and SAMS divisions in the Stockholm
region
20
Illustrating the 100m and 500m grid systems.
Example Data from the Tensta housing estate,
Stockholm city.
21
What scale matters?

• Research Q
• to what extent individual social mobility of
  adults is influenced by individual and
  neighbourhood characteristics, with a special
  focus on various levels of scale and various
  definitions of area compositions.

22
What scale matters? (2)

• Proposition It is reasonable to assume that if
  endogenous neighbourhood effects are in
  operation, such effects would be greater in the
  immediate surrounding of an individual and they
  would decrease as the size of the unit increases.
  However, for correlated effects it is more
  difficult to hypothesize which level would be the
  most important and the spatiality can also be
  expected to vary according to which outcome we
  decide to study.

23
What scale matters? (3)

• In this study we (Andersson Musterd 2006)
  operationalized neighbourhood at four spatial
  scales, running from the municipality, over an
  officially existing neighbourhood definition
  (SAMS) to coordinate-based bespoke neighbourhoods
  (500m, 100m). Using multivariate statistical
  techniques on employment and income development
  1995-2002 for all inhabitants residing in
  Swedens three largest urban regions, controlling
  for a wide variety of personal and household
  characteristics, we were able to confirm our
  basic hypothesis that contextual effects on
  labour market performance are strongest at the
  very local level and non-existent or weak at the
  municipal level.

24
What scale matters?
25
Distribution of the level of unemployment over
the 100m by 100m units, SAMS, 500m by 500m units,
and municipalities in metropolitan Sweden 1999.
26
What time matters?

• Duration of exposure
• Duration of effects
• Lagged effects
• Studies planned by not done so far.

27
Problematizing exposure timeNeighbourhood
movers and stayers

• Examples from two adjacent Stockholm
  neighbourhoods/districts
• Analyses of the 1990 cohort of neighbourhood
  residents
• and a note on selective migration in distressed
  urban neighbourhoods

28
Tensta A large housing Estate built around
1970 Pop. 15,600 in 1990.
Spånga, an older area, dominated by single
housing and home ownership. Pop. 6,600 in 1990
29
Neighbourhood staying frequencies 1990-2004 for
the 1990 population of Spånga in Stockholm city
(cohort size N6617 in 1990 2537 in 2004).
30
Neighbourhood staying frequencies 1990-2004 for
the 1990 population of Tensta in Stockholm city
(N 15567 in 1990 4206 in 2004).
31
Neighbourhood staying frequencies 1990-2004 for
the 1990 population of Tensta in Stockholm city
(by one year age groups)
Cohort sizes in 1990 age 5 271 persons, age
15 212, age 25 291, age 35 285, age 45 220,
age 55 102. Source GeoSweden 2004, Institute
for Housing Urban Research, Uppsala university.
32
Selective migration reproduces the positionof
poor and less attractive neighbourhoods
Employment frequencies by age in 1995 for people
staying in, moving into, and out of distressed
Stockholm neighbourhoods 1990-1995. (Age 20-64 in
1990)
Source Andersson Bråmå (2004)
33
Conclusions

• common policy thrust toward neighbourhood
  social mixing must be seen as based more on faith
  than fact (Galster 2007, Europ. J. of Housing
  Policy, 71, p. 35)
• If George Galster is correct we have a lot of
  work to do.
• Not all countries have data that allow for large
  scale longitudinal studies of neighbourhoods and
  neighbourhood effects but a wider European
  research agenda should also include systematic
  studies of the black box of neighbourhood
  effects, i.e. exploring what sorts of social
  externality processes actually are occurring in
  their nations neighbourhoods (ibid.) Studies of
  different kinds are needed to fill this research
  gap.

34
A research programme on neighbourhood mix and
neighbourhood effects
Global, National and Urban Contexts
The Micro Structure of the Housing Stock
(neighbourhoods composition in terms of tenure
and housing types)
(1)
Social and Ethnic composition of neighbourhoods
(2)
(3)
Social interaction Effects on attitudes and
behaviour
Social opportunities
(4)
35

• Thanks for your attention.
• End.