Professional Documents
Culture Documents
Choosing Neighborhoods: Residential Mobility and Neighborhood Careers
Choosing Neighborhoods: Residential Mobility and Neighborhood Careers
NEIGHBORHOOD CAREERS
ABSTRACT
There has been a long and ongoing debate in the housing choice literature about whether
and how households take neighborhood into account in the choice process, and in turn
how the neighborhood choices affect family and household quality of life. In the past
decade, neighborhood effects and neighborhood implications of residential choice have
taken on a greater role in discussions of residential mobility and residential satisfaction.
Overall, the literature on housing choice has emphasized the characteristics of the
dwelling, in contrast to the role of neighborhood as the major factor in the choice process.
But even if households mostly move house to make gains in the size and quality of the
dwelling, they may at the same time succeed in moving into neighborhoods with a higher
socio-economic status, or environments with a higher physical quality and more
greenspace. Moreover, the moves may be intended to improve neighborhood quality in
the first place, instead of to gain more housing space or quality. The present paper
examines the patterns of neighborhood choice and evaluates the relative role of
neighborhood in the residential mobility process. In the Netherlands a significant
percentage of residential moves leads to gains in the socio-economic status of the
neighborhood and the amount of greenspace in the new location. Households not only
move up in the housing hierarchy, but also fairly consistently make a neighborhood
career in the residential mobility process.
INTRODUCTION
Gaining access to decent housing and pleasant communities with high quality facilities
and environments is fundamental in the housing choice process. All households,
American or European, want to live in good quality houses and in neighborhoods which
feel safe and provide the amenities which ameliorate the stresses of urban living. Parks
and other greenspaces are an important part of making environments attractive, but so too
is the presence of friendly neighbors and the absence of crime and deteriorating physical
structures.
Although there is substantial work on the way in which households make their
choices in the housing market, the relative role of elements of the house and elements of
the neighborhood are still only partly understood. Recent research has suggested that the
neighborhood is an important factor in housing satisfaction and new research shows that
neighborhood does matter for a number of social outcomes. At the same time we still do
not have a good grasp of how households improve the status and environmental quality of
the community or neighborhood when they move. Even though an old adage in the real
estate community, (what are the three most important factors about choosing a houselocation, location, and location), stresses neighborhood and community, in most studies,
aspects of the house continue to show up as the factors most influencing the process of
residential mobility.
Focusing on neighborhoods and communities and their role in residential relocation
resonates with the increasing focus on place both in geography and planning. For
instance, in the most recent Housing Memorandum for the Netherlands (MVROM,
2000a) household access to the most desirable residential environment is one of the main
themes. The same concern is true in the US in the debates about new urbanism and smart
growth (Porter, 2000). Place is of increasing interest for its potential independent effect
on a variety of social outcomes. Thus, changing neighborhoods can be an important
element of increasing or decreasing social status. Moreover, people are also increasingly
concerned about the environmental factors which are of course integral with
neighborhood design and density and believed to affect health and well-being (Takano et
al., 2002).
Thus, the central focus of the current paper is to attack directly the issue of the nature
of change in neighborhood characteristics, in the residential move. It is well known from
the literature that households predominantly make upwards moves in the housing
hierarchy. Is this upwards pattern also true for neighborhood trajectories when people
move? And do gains in housing quality coincide with gains in neighborhood quality, or
are both partly independent in the residential move?
In particular we wish to examine the actual moves through a neighborhood status
hierarchy in two respects. First we examine whether households in general and in which
circumstances in particular, make an upward (or downward) move between
neighborhoods of different socio-economic status. Secondly we will examine to what
extent residential moves are also moves from neighborhoods with few environmental
amenities, like parks and greenery, towards locations well endowed with these amenities.
THE RESEARCH CONTEXT PREVIOUS WORK
Why neighborhood matters
The literature on the importance of housing is extensive (Rohe et al., 2000, 2001). As
households grow and age they move through a series of different housing types. The
housing career, as it is called, is short hand for the successively larger (and more
costly) housing that a family accesses as the size of the household grows and as the
resources to purchase housing increases (Clark et al., 2003). The housing career, then, is
a series of different dwellings that the household selects to meet their space and other
needs. But the locational specificity of housing means that the household acquires not just
the physical unit but a neighborhood and a set of services and qualities attached to that
neighborhood. Because houses come with a broader community and a narrower
neighborhood, there is increasing interest in how the spatial location influences housing
outcomes and whether households consistently move to better neighborhoods in their
search for a high quality of life (Sirgy and Cornwell, 2002).
There is a growing body of literature that suggests that neighborhood conditions do
play a role in shaping individual outcomes (Ellen and Turner, 1997). Obviously, people
want to live in good neighborhoods and to have their children grow up in safe
environments. Almost, without exception, there is a perception that growing up in a
bad neighborhood will likely reduce the life chances of those children (Musterd and
Ostendorf, 1998). The reviews of neighborhood effects have tended to stress the effects
of poverty and low income neighborhoods and concluded that indeed there are
neighborhood effects on such outcomes as childhood achievement (Duncan, BrooksGunn, and Klebanov, 1994); victimization in unsafe neighborhoods (Sampson, et al,
1997), and of course access to the labor market (Ihlanfeldt and Sjoquist, 1989). This
literature mostly refers to the socio-economic characteristics of neighborhood.
But neighborhoods can of course be perceived from a variety of perspectives beyond
those related to socio-economic status (Martin, 2003; Webster, 2003). Attention to the
environmental quality of neighborhoods and the influence this may have on the well
being and health of residents is increasing (Jackson, 2003; Stokols et al., 2003). Some
studies have evaluated the specific role and value of greenspace in cities and its impacts
on residential well-being (Pauleit, 2003). One study in the Netherlands found that the
amount of greenspace in the living environment effects (self-reported) health in a positive
way (De Vries et al., 2003). Rural aspects of neighborhoods are also very much valued in
the residential choice patterns of movers in The Netherlands as Van Dam, Heins and
Elbersen (2002) have recently demonstrated. Thus, it is not surprising that both
relationships with neighbors and environmental quality of the neighborhood are
frequently identified as important elements in residential satisfaction, the propensity to
move and the choice of a house and neighborhood (Amrigo and Aragons, 1997;
Hooimeijer and Van Ham, 2000).
Research also shows that the neighborhood effects on quality of life, residential
satisfaction and in the residential choice process are intertwined with family socioeconomic characteristics. In general, higher status socio-economic groups experience
more residential satisfaction than lower status socio-economic groups. Clearly there are
complex links between social position and residential environment. So in analyzing
neighborhood choice patterns it is crucial to take income into consideration, because it
may be expected to be a dominant factor not only in the dwelling career, but also in the
neighborhood career.
Homeowners are more likely to have higher residential satisfaction than renters,
perhaps a function of the overall higher quality of owner over rental dwellings. A
specific study of homeownership effects suggested that individuals are willing to pay
more to live in neighborhoods with owners rather than renters (Coulson, et al 2003). The
study shows that even after a variety of controls (physical characteristics of the dwelling
and neighborhood characteristics including ethnicity) higher rates of homeownership in
neighborhoods are accompanied by higher housing prices within the neighborhoods. In
effect there are significant externalities to homeownership. Or more simply,
neighborhoods with many more homeowners are more desirable neighborhoods.
Homeowners are also more likely to participate in local political activities, have
relatively high rates of voting and participate in local volunteer actions. Because
homeowners move less frequently than renters, this stability may lead to stronger local
social networks. An extensive discussion of the neighborhood effects literature is
contained in Rohe, Van Zandt, and McCarthy (2000, 2001) and Dietz (2002). For our
analysis of residential moves between neighborhood of different quality, it is important to
conclude from this literature that income and tenure choice may be expected to be major
elements in gains in socio-economic and the environmental quality of neighborhoods.
2. To what extent are the gains in housing housing quality and neighborhood socioeconomic status or neighborhood environmental quality coincident or independent of
housing gains?
We know from the literature that gains in housing quality predominate over gains in
neighborhood quality, because dwelling characteristics are most frequently stated as
a reason to move. But at the same time it can be expected that neighborhood quality
improves in conjunction with moving up the housing hierarchy for many households.
And because neighborhood considerations are identified by some 20 per cent of
households as a reason to move in the Netherlands, we also expect some households
to move to better neighborhoods without an accompanying gain in housing quality.
Speculatively we expect such moves most frequently in the rental sector, because in
the Netherlands this sector is mostly public housing and highly regulated and it is not
easy to make gains in housing quality with a move in this sector. And public housing
is mostly found in central cities where the desire to move for neighborhood reasons
is most frequently mentioned (MVROM, 2003).
3. What are the variables which explain successful gains in neighborhood quality, house
quality or both together?
As in the movement of households up the housing ladder we expect that income,
household type and change, tenure position and change, and the regional structure of
the housing inventory in the Netherlands, will play important roles. For instance
neighbourhoods with abundant greenspace are less available in the large cities in the
Netherlands than in middle sized cities (MVROM, 2000a). And rental housing is 71
per cent of the stock in the four largest cities, 53 per cent in middle sized cities and
only 32 per cent in municipalities with less than 20,000 inhabitants. It has been
demonstrated that this influences housing choice (Dieleman and Wallet, 2003), but
may also be expected to influence the possibilities of making gains in neighborhood
quality in the residential move.
DATA AND NEIGHBORHOOD TYPOLOGIES
The data for our analyses come from three sources. The first and most important source is
the Netherlands National Housing Survey 1998 (Woning Behoefte Onderzoek 1998;
WBO 98; MVROM, 2000b). The WBO is a large sample survey of Dutch households
conducted every four years. The 1998 sample consisted of approximately 70,000
households. By weighting the cases the sample gives a representative picture of the
housing market position of the Dutch population. The WBO has information on a wide
variety of housing characteristics, including data on persons and households and their
characteristics, housing situation and satisfaction and residential mobility behavior. From
this file we selected all households that moved residence over the 1994-1998 period, and
who lived in an independent housing situation before and after the move. (We excluded
union formation and moves from the parental home.) In total there are 32,908 cases.
Characteristics of the previous and present dwelling, including tenure, price and size, are
available and were used in the analysis. Household characteristics, income and household
position and change are available for the year of the survey 1998; change in household
composition before and after the move is also available. We used information on the
geographical location of the house both before and after a residential move to identify the
neighborhood characteristics of the households before and after they moved. However,
detailed information on the previous location is not always available.
We used two different sources to classify neighborhoods by socio-economic status
and by physical environment.1 The classification of neighborhoods by socio-economic
status is based on a typology developed by the Social Cultural Planning Bureau (SCP,
1998). They developed this typology for nearly all 4000 four-digit postal zones in the
Netherlands. The postal codes have on average about 4000 inhabitants, but there are
substantial variations in size. Data on income, unemployment and educational level were
used to develop a factor score for the social status of the neighborhood (we use the
abbreviation SES throughout the text). In the first part of the analysis we ordered the
postal zones on the bases of the SES score, and then subdivided the scores into five
classes of approximately equal size (800 postal zones) from low (1) to high (5) SES, and
constructed origin-destination matrices for the households that had moved across these
classes. Thus, in SES category 1 the neighborhoods are relatively poor neighborhoods,
with relatively many unemployed and many inhabitants with low levels of education.
Neighborhoods in category 5 are at the other end of the socio-economic status scale and
relatively wealthy. In the second and third part of the analysis we used the factor scores
themselves rather than a classification into classes.
For the classification of neighborhoods by physical environment we relied on the
elaborate typology developed by Harts et al. (2000). They used a detailed 250 by 250
meter grid pattern for the whole Netherlands, and land use data for these grids for the
1990-1996 period, as the basis for their typology. Their aim was mainly to differentiate
urban land use with respect to density of land use and mixing or specialization of land
uses. Each grid was assigned to one of 15 categories of urban land use or to one category
of rural land use. We simplified this typology from 15 types to 8 types, which we
interpret as an indication of the quality of the physical environment. These eight types of
land uses are treated as an ordinal scale indicating quality of the physical environment
from low to high (Table 1).
Highly urban central neigborhoods (category 1, Table 1) have a high land use density
and a mix of services, retail and housing. These areas are predominantly found in the
historical cores of the large cities and in a limited number of areas more at the periphery
of these cities. The category also includes urban neighbourhoods of somwhat lower
density mostly located in the cores of smaller cities like Hilversum, Heerenveen,
Kerkrade etc. In categories 2-4, the residential function dominates as the description of
1
Neighborhood is both a social and a spatial concept and the concept of community is often used to
describe the social dimension although in much of the literature neighborhood and community are only
loosely defined (Martin, 2003). Indeed, there is little agreement about the differences between the concepts
of neighborhood and community, and just what the spatial structure of a neighborhood or a community is,
or how it relates to standard data collection techniques, especially the role of census tracts in the US. With
the ability to geo-code areas it is possible to assemble units for specific analysis, as we do in this paper but
we are still some distance from any agreed spatial or social definition. We have in this paper to fall back on
statistical definitions of areas, but we raise this issue to indicate that it is still unresolved and we are aware
of it.
for movers in the highest status areas it is a matter of maintaining their privileged
position.
Important insights are to be gained by contrasting the lowest and highest income
households. We summarize the results as graphs of within class, upward and downward
moves for the five classes. There are large differences in the results. Higher income
households are able to maintain their higher status neighborhoods (nearly 60 per cent
move within the category) and those in the lower status neighborhoods are able to make
substantial gains (nearly 62 per cent move to higher status neighborhoods). In contrast,
lowest income households are less able to maintain their high status neighborhoods
(Figure 1). Only 38 per cent of the lowest income group in the lowest status
neighborhoods are able to make gains in status. As expected income is a an important
factor in the ability to change the class of neighborhoods status.
Tenure provides another perspective on the choices and movement patterns across
SES status (Figure 2). Clearly, owners are advantaged. Those who move within the
owner sector, even if originally in lower status neighborhoods are able to maintain or
make significant gains in neighborhood socio-economic status. Households who move
from rent to own also predominantly make gains in neighborhood status. This is
especially true for those in the lower status origin areas. We might hypothesize that
moves within ownership, and from renting to ownership, might well be motivated by the
wish to live in a better neighborhood and not simply the wish to move into more
luxurious housing.
The geographic structure of the Netherlands, particularly the division between
moving within the Randstad cities and moving within cities outside the Randstad is an
important context for household moves across neighborhoods (Figure 3). In the central
cities of the Randstad (especially Amsterdam and Rotterdam) the rental stock is
dominant, it makes up nearly 80 per cent of the housing inventory. In the central cities
outside the Randstad the variation in housing stock in term of tenure and quality and also
in terms of neighborhood economic status is much larger (Dieleman & Wallet, 2003). In
the Randstad it is therefore much more difficult to make a neighborhood career in socioeconomic status than outside Randstad Holland, by a move within the central city. As a
result, within the Randstad one needs to move to the suburbs to live in a good
neighborhood, while elsewhere one can find such residential environments within the
city. In fact with a move within the central cities of the Randstad it is quite likely that a
household will loose neighborhood status when it relocates. The real contrasts are
between the graphs that show movements from central city to suburbs and within the
suburbs (Figure 3). Households who move from a central city to one of its suburbs make
extremely large gains in neighborhood status especially if they were originally living in
middle and lower status communities. For example, nearly 80 per cent of households in
category 3 make a gain in neighborhood socio-economic status with their move.
Similarly, households who move within the suburbs who are already in higher status
neighborhoods are able to maintain that status more than 70 per cent do so.
In summary, the analysis of moves of households within and between neighborhoods
by socio-economic status, shows a much wider variation in upwards and downwards
moves, than we originally expected. The dominant trend is moves from lower to higher
status neighborhoods, but the variation between groups of households in lower and higher
income brackets, those in renting and owning and those living within or outside the
Randstad are dramatic. We will return to the influence of these variables on making a
neighborhood change later in the analysis.
1b. Transitions across Physical Environments2
Moves across neighborhoods classified by quality of the physical environment and
amount of greenspace (Table 1) show a wide variety in upward and downward moves
(Figure 4a). The general trend towards neighborhoods with lower density and with more
greenspace, what we interpret as an upward move in the quality of the physical
environment, is quite evident. Those previously living in more compact and densely built
neighborhoods with mixed land use (categories 1-4) in the main move to neighborhoods
with more open space, especially to areas in category 6, lower density predominantly
residential areas. And those already living in such areas with abundant greenspace
(categories 6 and 8 rural), have a very low prosperity to move to the denser and more
mixed land use neighborhoods closer to the city centers (categories 1-4). From these
patterns one can certainly identify the low density urban residential neighborhoods and
the countryside as the most preferred neighborhoods in terms of physical environment in
the Dutch situation similar to low density suburban environments in US metropolitan
areas (cf. MVROM, 2000b).
But even if the upward trend in the physical quality of the neighborhood in the
process of residential mobility is evident, upward and downward moves between all the
categories are frequent (Figure 4a). This is certainly true if one takes characteristics of
households and their dwelling and location in the Netherlands into account (Figures 4b
and c, Figure 5). For ease of presentation, we simplify the eight categories of physical
environment (Fig. 4a) into four dominant types in the rest of the figures. We collapsed 1
and 2 into one group and refer to it as center urban, and merged 3 and 4 called residential
urban in this part of the paper. We eliminated 5 and 7, because these are quite
heterogeneous and have relatively small numbers of residential moves (therefore
percentages in the reduced figures do not add up to 100). For convenience we refer to 6
as suburban and 8 as open space. The collapsed categories by income provide a story
which parallels that for the SES analysis by income (Figure 4b and c). If we assume that
the low density/open space physical environments are the most desired, then the
increased ability to access the most open space environments is a reflection of greater
income. Income, in this case equals access. The most open space neighborhoods are
retained by 57 per cent of high income earners and those higher income households who
are in the center urban and residential urban areas make significant moves to suburban
and open space neighborhoods.
The interaction of physical environment and tenure plays a large role in the move to
higher quality neighborhoods (Figure 5). Movers within ownership especially, but also
from rent to own have very high proportions who moved to the most desirable
neighborhoods, those categorized here as suburban and open space (categories 6 and 8).
In the Netherlands renting coincides with central and dense residential urban
environments and owning with suburban and open space areas. Once households have
attained home ownership they tend to remain in low density areas with relatively high
levels of open space. Only very small numbers of moves are from suburban and open
2
Recall that the physical neighborhood, based on more detailed post codes, is smaller than the SES
neighborhoods.
10
space environments to more urban settings and only for those moving within the rental
sector (Figure 5).
Regional structures also play a role in physical environmental choices. Cities in the
Randstad have less open space and more neighborhoods of higher density than cities
elsewhere. This influences movement across neighborhoods of different environmental
quality. Those moving within cities outside the Randstad retain and gain much more
quality of the physical environment than those moving within cities within the Randstad.
Central city relocations to one of its suburbs do increase the quality of the physical
environment greatly (though we only have data for the most dense physical
environments). Moves within the suburbs of the same central city have of course large
proportions who are able to maintain their current physical environment or improve it
(Figure 6d). It appears that they are able to use the internal suburban movement to change
their physical environment if you will, to improve it by attaining lower densities and
more open space.
Summarizing the gains in environmental quality with the residential move reveals a
remarkable similarity in comparison with the patterns of gains and losses in socioeconomic status of the neighborhood. This is the more remarkable because the
measurements of socio-economic status and environmental quality of neighborhood
follow quite different lines both in terms of what a neighborhood is and how quality is
defined. In both cases we demonstrated a predominant upward pattern of moves in terms
of neighborhood quality as we expected from the literature. But the variations are wide if
one breaks the moves down into those for lower and higher income households, by tenure
structure and by location within the Netherlands. 3
2.
Trading houses and neighborhoods
Thus far we have shown the way in which households with varying characteristics,
defined by income and tenure, make transitions within the socio-economic and physical
structure of neighborhoods in the rich urban structure of the Netherlands. Now, we add to
these results by directly examining how much neighborhood plays a role in those choices,
the second central theme of this paper. To what extent and in what contexts do
households trade-off quality and tenure of the house and socio-economic status and
physical structure of the neighborhood in their relocation behavior? Is there a clearly
separate role for neighborhood quality in the process of residential mobility or do gains in
neighborhood quality always coincide with making an upward dwelling career?
To evaluate the separate role of neighborhood in the mobility process we examined
whether or not there were gains in neighborhood status improvements, or in both house
and neighborhood, or no gains in the quality of either before and after the move. Gains
were measured by comparing the factorscore for each household at the previous and the
current locations to assess whether or not there was a change in ses neighborhood. A gain
was defined as: new factorscore minus old factorscore greater than .25, which is a real
and clear improvement in SES. A gain in physical environment was more naturally defined as moving upwards one or more categories in our ordinal typology of physical
3
We also explored the role of type of household and of change in size of the family in a comparable way.
The resulting patterns are not transparent and we do not present the results in the descriptive results.
However, in the regression models we examine the role of these variables.
11
Recall that rental housing is mostly social housing with relatively small variations in housing quality.
12
status of the neighborhood. This is because in most cities in The Netherlands income
differences between central cities and surrounding suburban communities are not great so
such a move does not necessarily result in a gain in socio-economic status of the
neighborhood (Dieleman and Wallet, 2003). But a central city to suburb move certainly
will result in getting into a lower density and greener environment, especially in the
Netherlands where a policy of compact city (re)-development has been pursued for
decades and central urban residential neighborhoods are very dense. In cities outside the
Randstad there are more neighborhoods with open space and greenery within the central
parts of cities, and moves within the city can therefore also result in gains in physical
environment of the destination.
To reiterate, the result of this evaluation of how housing choices can generate gains
in neighborhood quality stresses that while households look to make gains in their
housing quality we can infer from the revealed preferences that they are also evaluating
and trading up the neighborhoods in which they live. Overall, nearly 25 per cent of all
households who relocate make gains in their neighborhood socio-economic status or
housing quality and neighborhood status in combination. For the environmental
characteristics of neighborhood these gains are made in over 35 per cent of the moves.
The latter of course is a direct outcome of the general association between housing
quality and neighborhood quality, but 16 per cent of all moves result in a better
environmental quality of the neighborhood without gains in dwelling quality as we
defined it. The percentages of moves that result in lesser density and more greenspace are
much higher than we expected. In a recent survey in the Netherlands for instance only 21
per cent of the movers mentioned the character of the present neighborhood as a main
reason to move (MVROM, 2003).
3.
Multinomial models of change across neighborhoods
In the previous descriptive analyses the focus was on gains in neighborhood quality, by
the characteristics of the household and the dwelling and the geographical context in
which the residential move took place. Because these separate indicators are of course
interrelated we extend the analyses using a logit model to disentangle the relative
contribution of the factors which influence the likelihood of households making an
improvement in housing or neighborhood quality. We present models separately for
neighborhoods classified by socio-economic status (Table 4) and environmental quality
(Table 5)5. The dependent variable of the multinomial logistic regression model is
defined as making gains in dwelling quality only, in neighborhood quality only, in both
dwelling and neighborhood or making no improvements with the move. The last
category, no improvement, is taken as the reference category. So the first model in Table
4, improvement in both housing and neighborhood quality is in comparison to no
improvement in either. As independent variables we used income and we reintroduced
household structure and changes in household composition before and after the move.
Also the geographical structure of the Netherlands is used again as an explanatory factor
of gains in neighborhood quality. For each of the independent variables the last category
is taken as the reference category. For income this reference category is above 1.5
modal for instance. All models have acceptable levels of fit and explanatory power.
5
We present the significance level of the un-weighted model, because in the weighted model the number of
cases is so high that calculating significance levels becomes irrelevant.
13
Because the emphasis in this paper is on the role of neighborhood quality in the main
we emphasize the two models that refer to neighborhood, and pay less attention to gain
in dwelling quality only. For the model on gains in socio-economic status of
neighborhood only (Table 4), two parameters are important, those for one person
households and for the Randstad. One person households, probably many older
households, make gains in socio-economic status of the neighborhood only, more
frequently than others. These households are very clearly more concerned about the
socio-economic status of neighborhoods than others, if these revealed preferences may be
interpreted in that way. Families with or without children in contrast make more gains in
dwelling quality only. In the Randstad gains in socio-economic status of neighborhood
only are more frequent than outside the Randstad. Households maybe more aware of
neighborhood differences in the large urban agglomerations and seek to improve the
neighborhood in which they live.
When one looks at the model for improvement of both housing quality and socioeconomic status of neighborhood as compared to no gains in either the factors usually
encountered in the housing choice literature re-emerge. Income and household structure
and change dominate the models, while other indicators have low significance (Clark et
al., 2003). The model is quite comparable to gains in house quality only. Higher incomes
improve housing and neighborhood quality more than lower incomes, and larger and
expanding households move up the housing and neighborhood hierarchy more than
others. So it looks as if combined improvements of house and neighborhood in the
mobility process are quite similar to moves where gains in housing quality drive the
move; while moves where only neighborhood status is improved seem to be quite
distinct.
The logit model referring to improvements in the environmental quality of
neighborhoods (Table 5) is somewhat comparable to the previous model for SES status of
neighborhoods. In the model for improvement of physical environment only, again
household type and Randstad versus non-Randstad are dominant factors, as in the SES
model. Families with children make more moves to greener and more open
neighborhoods, which makes intuitive sense. And also for the quality of the physical
environment the Randstad has larger contrasts than elsewhere in the country and gains in
quality of the environment only are therefore more frequent. Moreover, a central city to a
suburb move directly changes the amount of greenspace in the neighborhood.
The models for improvement of physical environment and housing both and for
housing improvement only, are quite similar to the results for the SES models. Income,
household structure and metropolitan geographical structure all influence gains in
housing and neighborhood in the expected way, and the results mainly substantiate what
we concluded in earlier sections of this paper.
CONCLUSIONS AND OBSERVATIONS
The current paper poses three questions is there progress in neighborhood quality with
mobility, is the focus in the mobility process mainly on improving the quality and size of
the dwelling or are there (also) major gains in status and amount of greenspace of the
neighborhood, and what variables are the central explanatory factors in the trade-off of
house and neighborhood.
14
15
REFERENCES
Amrigo, M. and J.I. Aragons (1997) A theoretical and methodological approach to the
study of residential satisfaction. Journal of Environmental Psychology 17, 47-57
Briggs, X. de Souza (1997) Moving up versus moving out: neighborhood effects in
housing mobility programs. Housing Policy Debate 8, 195-234
Clark, W.A.V. and Dieleman, F. (1996) Households and housing: Choices and outcomes
in the housing market . Rutgers, Center for Urban Policy Research
Clark, W.A.V., M.C. Deurloo & F.M. Dieleman (2003) Housing Careers in the United
States, 1968-93: Modelling the sequencing of housing states. Urban Studies, 40 (1), 143160
Coulson, N.E., Hwang, S. and Imai, S. (2003) The value of owner occupation in
neighborhoods. Journal of Housing Research 13, 153-174
Dam, F. van, S. Heins & B.S. Elbersen (2002), Lay discourses of the rural and stated and
revealed preferences for rural living. Some evidence of the existence of a rural idyll in the
Netherlands. Journal of Rural Studies 18: 461-476
Dieleman, F.M. and C. Wallet (2003) Income differences between central cities and
suburbs in Dutch urban regions. Journal of Economic and Social Geography, 94, 265-275
Dietz, R. (2002) The estimation of neighborhood effects in the social sciences; an
interdisciplinary approach, Social Science Research 31, 539-575
Duncan, G. Brooks-Gunn, J. and Klebanov, P. (1994) Economic Deprivation and early
childhood development, Child Development 65, 296-318
Ellen, I.G. and Turner, M. (1997) Does neighborhood matter? Assessing recent evidence.
Housing Policy Debate 8, 833-866
Galster G. (2003) Investigating behavioral impacts of poor neighborhoods: Towards new
data and analytic strategies. Housing Studies 18, 813-914
Greenberg, M. (1999) Improving neighborhood quality: A hierarchy of needs, Housing
Policy Debate 10, 601-624
Harts, J.J., C. Maat and M. Zeijlmans van Emmichoven (1996) Monitoring stedelijke
milieus, menging en dichtheid. Stedelijke en regionale verkenningen 23, Delft, Delft
University Press.
16
17
Rohe, W.,Van Zandt, S. and McCarthy G. (2000) The social benefits and costs of
homeownership. Research Institute for Housing America, Working Paper 00-01
Rohe, W.,Van Zandt, S. and McCarthy G. (2001) The economic benefits and costs of
homeownership. Research Institute for Housing America, Working Paper 01-02
Sampson, R., Raudenbush, S. and Earls, F. (1997) Neighborhoods and violent crime: A
multilevel study of collective efficacy. Science 277, 918-924
SCP (1998) Van hoog naar laag; van laag naar hoog. De sociaal-ruimtelijke ontwikkeling
van wijken tussen 1971 en 1995. The Hague: Sociaal en Culturaal Planbureau
Sirgy, M. J. and Cornwell, T. (2002) How neighborhood features affect quality of life.
Social Indicators Research 59, 79-114
Stokols, D. Grzywacz, J. McMahan, S. and Phillips, K. (2003) Increasing the health
promotive capacity of human environments, American Journal of Health Promotion
Sept./Oct. 4-13.
Takano, T., Nakamura,K. and Watanabe, M. (2002) Urban residential
environments and senior citizens' longevity in megacity areas: the
importance of walkable green spaces. Journal of Epidemiology and Community
Health, 56, pp. 913-918.
Vries, de S., R.A. Verheij, P.P. Groenewegen & P. Spreeuwenberg (2003) Natural
environments - healthy environments? An exploratory analysis of the relationship
between greenspace and health. Environment and Planning A, volume 35, 1717-1731
Webster, C. (2003) The Nature of the Neighbourhood. Urban Studies 40, 2591-2612
18
Table 1
Type
1. Center, highly urban + Center, urban
2. Residential, highly urban
3. Residential, urban
4. Residential
5. Center, low urban + Residential and employment
6. Residential, low urban + Residential and greenspace
7. Service/retail concentration + Employment and greenspace
+ Greenspace and sports area
8. Open space/agricultural
Total
Source: Harts et al., 2000; WBO 98.
19
1428
1732
1843
2788
1884
6425
1043
1969
2463
2350
3832
2752
9768
1850
3767
20,910
7460
32,444
Table 2
Moving category
Only
dwelling
Only
neighborhood
Both
No
improvement
Total %
Al movers
35.0
13.7
10.4
40.9
100.0
Income
Above 1.5 modal
Modal to 1.5 modal
Minimum to modal
Below minimum
45.6
39.3
22.4
14.7
10.8
12.5
17.4
19.2
13.7
10.5
7.6
5.0
29.9
37.7
52.6
61.1
100.0
100.0
100.0
100.0
Tenure
Within rental sector
Within owner sector
From rent to own
21.4
28.3
75.8
19.3
15.8
0.0
5.9
7.3
24.2
53.4
48.6
0.0
100.0
100.0
100.0
Geographical context
Within central city Randstad
Within central city not
Randstad
29.6
35.9
15.0
12.3
9.7
10.0
45.7
41.8
100.0
100.0
Within suburbs
Central city to suburb
34.9
50.9
11.7
9.1
8.5
7.8
44.9
32.2
100.0
100.0
20
Table 3
Moving category
Only
dwelling
Only
neighborhood
All movers
25.4
16.3
20.4
37.9
100.0
Income
Above 1.5 modal
Modal to 1.5 modal
Minimum to modal
Below minimum
30.7
28.9
19.5
12.0
14.4
16.1
18.4
19.7
29.3
21.1
10.1
7.8
25.6
33.9
52.0
60.5
100.0
100.0
100.0
100.0
Tenure
Within rental sector
Within owner sector
From rent to own
16.4
20.0
53.7
21.0
22.2
0.0
10.8
16.1
46.3
51.8
41.7
0.0
100.0
100.0
100.0
Geographical context
Within central city Randstad
Within central city not
Randstad
24.5
24.6
17.0
15.4
15.4
22.1
43.1
37.9
100.0
100.0
Within suburbs
Central city to suburbs
24.7
11.9
17.0
27.2
19.1
46.2
39.2
14.7
100.0
100.0
21
Both
No
improvement
Total %
Table 4
Sig.
(unweighted
model)
Intercept
Exp(B)
Sig.
(unweighthed
model)
.001
Sig.
(unweighted
model)
.000
.000
Income
(ref. cat. = over 1.5 x
modal)
Below minimum
Minimum modal
Modal 1.5 x modal
.139
.245
.467
.000
.000
.000
.194
.244
.548
.000
.000
.000
.805
.785
.872
.135
.008
.033
.343
.572
.000
.000
.398
.701
.000
.000
.984
.806
.496
.006
.962
1.008
.949
.031
.816
1.353
.000
.000
1.275
.895
.004
.211
.931
.849
.895
.014
1.176
.004
.288
.162
1.824
1.018
.000
.310
.991
1.300
.106
.051
Household type
(ref. cat. = family without
children <18y)
One person
Family with children <18y
Randstad versus non
Randstad
Randstad
1.266
1.186
-2 Log
Likelihood
197973.029
Chi-Square
df
Sig.
Effect
32699.560
165273.469
24
.000
Intercept
Income
Change
household size
Household type
Randstad
.159
22
-2 Log Likelihood
of Reduced Model
Chi-Square
Df
Sig.
32699.560
110905.708
45675.336
.000
78206.148
12975.777
0
9
6
.000
.000
43279.757
34394.357
10580.197
1694.797
6
3
.000
.000
Table 5
Sig.
(unweighted
model)
Intercept
Exp(B)
Sig.
(unweighthed
model)
.000
Sig.
(unweighted
model)
.000
.000
Income
(ref. cat. = over 1.5 x
modal)
Below minimum
Minimum modal
Modal 1.5 x modal
.144
.181
.498
.000
.000
.000
.225
.295
.589
.000
.000
.000
.975
.890
1.001
.826
.018
.332
.300
.712
.000
.000
.442
.716
.000
.000
1.041
1.050
.134
.863
Household type
(ref. cat. = family without
children <18y)
One person
Family with children <18y
.652
1.374
.000
.000
.748
1.404
.000
.000
.611
1.247
.000
.000
1.234
.000
.847
.003
1.341
.000
.000
.126
1.355
1.028
.713
.232
4.616
1.025
.000
.151
6.394
1.026
-2 Log
Likelihood
203153.903
Chi-Square
df
Sig.
Effect
30974.750
172179.153
24
.000
Intercept
Income
Change
household size
Household type
Randstad
.172
23
-2 Log Likelihood
of Reduced Model
Chi-Square
Df
Sig.
30974.750
108238.897
43388.371
.000
77264.147
12413.621
0
9
6
.000
.000
42323.767
37932.221
11349.017
6957.471
6
3
.000
.000
10.1
12.9
13.6
56.0
3,397
100
5.9
11.0
13.7
50.6
18.8
3,601
100
9.9
13.6
50.1
11.1
15.3
3,969
100
13.6
48.9
12.1
11.7
13.6
3,654
100
Lowest status1
51.5
15.5
12.0
8.0
13.0
3,314
100
2,553
3,469
3,979
49.9
56.5
29.8
29.0
3
2
17.7
62.0
7.3
Highest status 5
54.0
19.2
51.8
28.3
38.0
50.1
100
13.7
100
100
100
100
3,773
4,161 17,935
40.7
46.2
30.2
19.2
3
2
11.2
38.3
43.0
33.4
47.4
45.8
61.7
59.3
100
23.6
100
100
100
100
52.0
51.7
33.4
32.0
3
2
19.1
60.2
48.0
100
14.9
100
100
17.9
50.1
49.5
100
31.3
100
39.8
38.1
49.7
28.3
14.6
3
2
7.8
39.1
61.9
100
22.1
100
100
32.3
53.1
44.4
100
47.8
100
60.9
43.9
46.1
33.1
24.9
3
2
11.2
40.9
48.9
30.5
44.6
40.0
59.1
56.1
100
20.8
100
100
100
100
43.5
43.0
30.5
64.7
56.5
100
13.1
100
15.5
41.4
44.0
100
100
35.3
100
25.5
40.5
47.1
36.0
49.1
3
2
15.4
56.1
55.9
50.3
28.7
43.9
13.0
3
2
2.9
7.5
100
34.5
100
33.3
32.2
49.7
78.7
8.4
10.7
100
92.5
53.8
18.2
15.9
3
2
4.9
53.9
51.9
28.4
55.7
43.2
46.1
100
100
86.4
70.5
100
28.0
100
59.5
100
16.8
100
23.4
27.5
39.7
47.1
100
100
100
100
100
100
Destination
Low density/
more green space 8
1.8
0.8
1.4
4.5
7.2
23.6
5.1
55.7
100
6.6
3.7
2.8
12.1
9.9
29.3
15.2
20.5
100
3.2
1.9
2.5
11.5
9.3
45.5
6.5
19.6
100
4.8
1.4
5.1
12.1
22.1
32.7
4.6
17.2
100
5.1
2.7
8.0
27.5
7.9
30.7
5.6
12.5
100
6.3
12.2
26.5
14.3
7.5
23.7
4.3
5.3
100
7.1
40.4
19.9
9.1
4.9
13.5
2.2
3.0
100
High density/
less green-space 1
25.1
7.9
8.0
15.6
5.1
23.4
5.3
9.7
100
Origin
Destination
1+2
more
green-space 8
6
3+4
50.5
86.4
10.7
85.3
35.9
27.7
Destination
46.9
more
green-space 8
6
14.3
3+4 16.9
Less
green-space 1+2
52.0
46.7
26.0
41.6
89.6
3+4
93.6
Less
green-space 1+2
7.9
30.4
32.3
57.3
88.3
25.7
83.8
31
Origin
Origin
3+4
1+2
43.8
44.7
60.1
84.9
90.5
Destination
3+4
1+2
more
green-space 8
Origin
44.5
30.9
6
3+4
15.8
Less
green-space 1+2
50.8
43.1
87.6
7.6
86.0
47.5
47.6
89.7
26.3
91.6
40.8
Destination
3+4
1+2
more
green-space 8
26.2
9.6
60.9
87.1
29.3
82.2
43.3
Origin
3+4
Less
green-space 1+2
28.3
6.2
87.0
52.5
89.7
63.2
26.5
Destination
3+4
1+2
more
green-space 8
6
29.0
15.7
63.7
92.7
24.1
84.3
44.5
Origin
3+4
8.9
Less
green-space 1+2
30.6
27.0
46.8
60.4
82.7
91.0
Destination
3+4
1+2
more
green-space 8
6
Destination
81.4
56.4
18.6
100.0
1.3
93.4
35.7
Origin
3+4
1+2
more
green-space 8
59.8
33.6
35.5
95.3
8.7
86.7
44.4
Origin
3+4
26.5
Less
green-space 1+2
63.4
52.3
13.9
33.0
92.7
3+4
96.4
Less
green-space 1+2
7.7
36.6
48.5
27.9
52.0
Destination
3+4
1+2
50.7
14.3
79.2
more
green-space 8 Small n
6 14.2
Origin
3+4
7.4
Less
green-space 1+2
4.4
24.9
78.9
46.6
80.3
75.9
Destination
3+4
1+2
more
green-space 8
6
44.6 81.5
36.9
21.4
50.6
12.5 84.5
Origin
3+4
7.6
Less
1+2
green-space
27.7
42.2
36.5
61.8
86.3
89.5
84.1
88.6