Staying in the city or moving to the suburbs? Unravelling the moving behaviour of young families in the four big cities in the Netherlands

Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

attend school (Kulu & Steele, 2013). The growth of the suburbs in the Global North in the second part of the 20th century depended in the past for a large part on families moving from the city to the suburbs. But this clear relationship between cities and suburbs has changed, at least in some suburban areas (Tzaninis & Boterman, 2018).
The changing modes of production in the economy are an important driver for the growth of cities (Storper & Scott, 2009).
Cities have become wealthier, and many of them have undergone processes of gentrification. Gentrification is therefore an important element in understanding the changes postindustrial cities go through. The gentrification literature describes how the middle classes reclaim the city (Hamnett & Butler, 2013). Often, this reclaiming of the city is portrayed as a reaction to the monotony of the suburbs: 'How do we account for the movement into old neighbourhoods of, say, middle-class people who have the means to live where ever they choose, who by any demographic measure would two decades ago have settled in suburbs, and today seek the city?' (Caulfield, 1994, p. 133). The early gentrification literature spoke mainly about small households in specific professions (young single households, the creative sector), but later studies showed that the phenomenon was diversifying. Also, more middle-class families seem to become a growing phenomenon in the city (Boterman, Karsten, & Musterd, 2010). These gentrifying families express the need to stay in the city; they live a life that in their view would not be possible in a suburb (Karsten, 2007). This literature suggests that there is a change in residential preferences for middle-class families: from more suburban to more urban. But the question is whether there is really a change in residential preferences.
The literature on contemporary processes of urbanisation also recognises that cities are growing, but is less definite on the argument that this is caused by a change in residential preferences. It points to demographical transformations and argues that the growth of cities is primarily related to changes in the demographic and socio-economic composition of the population (Kabisch & Haase, 2011;Kabisch, Haase, & Haase, 2010). The diversification of household types and family formation due to the 'second demographic transition' (declining fertility rates, postponement of child-bearing and marriage, growing household numbers, smaller household sizes and nontraditional household structures) are all associated with a more urban preference (Buzar, Ogden, & Hall, 2005). But it also assumes that the preference of the traditional middle-class family with two cohabiting partners did not change and is still more suburban.
The literature on stated residential preferences also suggests a high degree of stability of preferences for urban or suburban living.
Residential preferences do vary through the life-course, but they are quite stable through time (Booi & Boterman, 2019;Fuguitt & Brown, 1990). Families still seem to have a strong preference for suburban living (De Groot, Van Dam, & Van Amsterdam, 2015). This begs the question whether there is a change in residential mobility for young families (households with two cohabiting partners and young children not older than 5 years old), do they less often choose to move to the municipalities in the region of the city and rather stay in the city? Or is the change mainly caused by a change in the demographic and socio-economic composition of young families? This paper addresses the following research questions: What is the demographic, socio-economic and socio-cultural composition of young families in the four big cities in the Netherlands and how does it change through time?
To what extent can the change in outmigration of young families to the municipalities in the region around the city be explained by changes in the composition of these young families in these cities?
What are the odds that a young family with specific demographic, socio-economic and socio-cultural characteristics moves from these cities to the municipalities in the region around the city and how has it changed through time?

| THEORETICAL FRAMEWORK
The life course approach is helpful to understand moving behaviour of families (Bailey, 2009;Clark & Lisowksi, 2018;Mulder, 1996). Residential moves are often related to household transitions. Young families are typically looking for better housing and neighbourhood conditions (Gambaro, Joshi, & Lupton, 2017). The timing is very precise, when families move it is often just before or just after the birth of the first child (Kulu & Steele, 2013). When the children are older, the mobility rate drops significantly. Young families not only look for larger homes that better suit their increasing family size (Dieleman, 2001) but also often a transition to home-ownership is made, typically before family-formation (Mulder, 2006).
The housing career of households is related to the job career (Clark, Deurloo, & Dieleman, 2003). People often try to match their own social status with the status of the neighbourhood (Clark, van Ham, & Coulter, 2014). The phase in which people start a family often coincides with the phase upward mobility is realised. A move to match the new housing needs as a family will therefore also be linked to a move to a neighbourhood that matches the (improved) social status. Schaake, Burgers, and Mulder (2014) show that this also is the case for ethnic minority groups, although in a lighter pace.
In terms of location many families still prefer suburban locations (De Groot, et al., 2015). An important part of that suburban ideal is associated with the geography of housing supply. A key reason to move to suburban locations is the availability of more spacious, more affordable, owner-occupied housing (Hayden, 2003). Yet not every young family can afford to buy a home. Lower-income or unstableincome families remain more dependent on the (social) rental sector, which is generally-so far-mainly found in more urban areas and in new towns. Moving from the city to the suburbs is therefore not only about housing preferences but also about housing constraints (Boterman, Manting, & Musterd, 2017). Many central cities in the Global North are becoming more expensive due to gentrification, and suburbs are getting both more socio-economically and ethnically diverse and may even gentrify (Paccoud & Mace, 2018). Most suburbs are no longer the monofunctional residential areas for middle-class families, as they are portrayed (Bourne, 1996;Keil, 2018). These variegated trends cause a shift of the geography of where affordable family housing can be found in metropolitan areas (Hochstenbach & Musterd, 2018).
The relationship between (changing) supply and residential preferences is however more complex. Stated residential preferences often reflect the actual possibilities households have (Clark & Dieleman, 1996). Furthermore, (underlying) preferences may come to the fore when new possibilities in housing arise (Storper & Manville, 2006). In other words, the different trade-offs that are made between price, tenure, location and other aspects of the home are affected by contextual changes at various levels. For instance, if more urban owner-occupied single dwellings are built, this might tip the balance for some families to prefer urban locations, making them to decide to rather stay in the city. The choice for staying in the city or moving to the municipalities around the city is therefore not only about residential location but also about housing supply in terms of price, size, tenure and other factors.

| Diversity of families
Young families in the city are not a homogeneous group. They are differentiated by their residential biographies, their preferences and their resources, affecting their moving behaviour. There is for instance quite some evidence that people when they want to settle down move back or near to the place they grew up (Feijten, Hooimeijer, & Mulder, 2008) or where their family lives (Blaauboer, 2011). The place of birth and where people grew up thus influences the choice of staying or leaving the city as a family. However, the longer people stay in one place, the less likely it is that they will move (Huff & Clark, 1978). City dwellers who originally came to the city for education or work typically stayed for quite some time (Butler & Hamnett, 2012;Tano, 2014). New urbanites may have become used to the urban lifestyle, invested in social networks and develop also symbolic and mental attachment to the city, making them inert and less likely to leave (Fischer & Malmberg, 2001). People who grew up in the city are expected to have even stronger attachment to the city, followed by people with a long duration-of-stay (albeit originally coming from elsewhere).
A growing group of young people come to the city for education or work. When they have children, not all of them leave for the suburbs, but some stay. This leads to a growing number of middle-class families in cities. This growth of middle-class families is often analysed in the context of gentrifying cities, as the settling down of 'gentrifier families' (Boterman, 2012;Boterman et al., 2010;Brun & Fagnani, 1994;Lilius, 2014). This research suggests that families live in the city because the proximity of work, daycare, shops and other daily amenities enables them to combine a family with a career. They express a strong attachment to the city and urban life and are willing to give up better housing quality, for example, live in a smaller house or stay renter, than they could have in the suburbs. There is quite some evidence that the cultural orientation of a household, expressed through the field of education or work, influences the social geography in metropolitan regions (Boterman et al., 2017). Households with a background in the social sciences or working in the creative sector seem to be more urban oriented. These specific fractions of the middle classes do not only command substantial cultural capital, they also often have a higher income, giving them relatively much choice on the housing market: they can choose to stay in the city or move to a more suburban location. But this higher income does not mean that they can afford everything; they still have to make tradeoffs (Boterman & Bridge, 2015;Bridge, 2003;Brun & Fagnani, 1994).
The economy of postindustrial cities in the global North increasingly has an international component. This is not only the case for so called 'Global Cities' (Sassen, 1991) but also for smaller noncapital cities (Bonneville, 1994). High-skilled workers move to these cities, also across borders. Expats become more numerous. Expats are often associated with small households that only live for a short term in a foreign country (Beaverstock, 2005). But some of them come as or eventually settle down as a family (Bontje, Musterd, & Sleutjes, 2017).
This new group of 'expat families' seem to have a lot in common with the 'gentrifier families', but their perspective on appropriate residential milieus may differ from Dutch families. Given the relatively small scale of the Amsterdam metropolitan region compared with their experiences from, for instance London, they may more easily move to the metropolitan region around Amsterdam. On the other hand, the strong presence of people from the same country as well as the associated amenities and shops may rather contribute to a central urban orientation (Beaverstock, 2005;Scott, 2006).
The composition of many Northern and Western European cities is influenced by their colonial history and their need for cheap labour in the 1960s and 1970s. During that period, many people from former colonies and guest workers came to Northern and Western Europe.
These families, here labelled 'ethnic minority families', often have a long history in the city. Contrary to the 'expat families', the parents from the main immigration countries have been in the city for many years. They came for a large part in the 60s and 70s of the last century and therefore have a specific demography and geography.
The first generation is mostly retired and generally past the family formation age. Demographically, the second generation is now in the family formation age. Originally, these groups were mainly concentrated in cities, but their geography is diversifying (Catney, 2016).
Although many still belong to the relatively weaker socio-economic groups, upward mobility now provides opportunities to purchasing a home in the suburbs and become owner-occupier (Hamnett & Butler, 2013). It has been demonstrated that the second generation is better educated and has better prospects on the labour and housing market than the first generation (Lelie, Crul, & Schneider, 2012). This could result in a suburbanisation trend of 'ethnic minority families'.
But other factors like the preference for living near to others from the same ethnic group or living in a more diverse neighbourhood can hamper them to move to a more suburban location (Andersen, 2016;Clark, 2009).
The last important group of families are those who are originally from the city, but are not of minority descent. These 'original city dweller families' may have a strong attachment to the city, yet they did not deliberately choose to live in the city. Historically, these 'original city dwellers' belonged largely to the working classes or lower middle classes. This group massively suburbanised in most of the 70s and 80s of the previous century. When they could afford it, they moved to a suburban setting to a single family dwelling. This was also motivated by the desire for a safe neighbourhood and good schools (Hinshaw & Allot, 1972). In the context of the gentrifying city, these groups, labelled as the 'ordinary people' (DeSena, 2006), are among those that face displacement. Given the diminishing size of the traditional white working classes, the question is does this group still constitute a substantial element of suburbanisation?

| Case study
In the past five decades, the population of the larger cities in the Netherlands has changed significantly (PBL, 2015). Changes that can also be seen in other postindustrial cities in the Global North. The three largest cities, Amsterdam, Rotterdam, The Hague and to a lesser extent also the fourth city, Utrecht, all have gone through a suburbanisation period where many city dwellers left these cities to live in the municipalities in the region around these cities. This period, roughly between 1960 and 1985, has been followed by a slow recovery in population size, and this growth accelerated after 2000 (PBL, 2015). In this suburbanisation period, the region around these cities still grew demographically, yet slowing down when the central cities recovered. In some suburban areas, policy makers are now even worried about population decline (PBL, 2015).
In this period, the composition of the population of the cities changed as well (PBL, 2015). Many families that left the cities during the suburbanisation period had no migration background and were often born and raised in the city. At the same time, the cities received a growing migrant population of guest workers (mainly from Turkey and Morocco) and from former colonies (mainly from Suriname and the Dutch Antilles). Cities with universities, especially Amsterdam and Utrecht, saw a growing number of students. In more recent years, the internationalisation of the economy lead to a rising share of expats (Savini, Boterman, van Gent, & Majoor, 2016). The Hague has a strong international component due to the presence of the international court of justice and many (related) nongovernmental organisations (NGOs) (PBL, 2015). Rotterdam kept a more working class profile, although it is diversifying (Nientied, 2018). These changes in the inflow of young people in cities ultimately lead to a change in the young families that are formed. The extent to which these changes in the population are manifest in the four largest cities of the Netherlands differs between the cities. These differences between the cities give the possibility to compare the outcomes and get a better grip on the influence of the changing composition on the residential mobility of young families from the city to the municipalities in the region around the city.
The focus is on young families that lived in the city and stayed within the metropolitan region of that city. When they move to the suburbs around the city, they often stay functionally tied to the city.
The moves young families make within a metropolitan region are often related to housing and neighbourhood quality. It is an expression of their residential preference and possibilities within a metropolitan region. Moves to locations further away are often more related to changes in jobs and/or working locations and are therefore left outside the scope of this article.
Moving behaviour is influenced by economic conditions, especially in economies where the housing market is highly related to the general economy. Periods of boom can lead to higher mobility rates, where periods of bust decrease mobility (Salvati & Carlucci, 2017), especially for owner-occupiers, making it more of concern for higher income households than lower income (Hochstenbach & Musterd, 2018). The economic crisis had a large impact on moving behaviour in the Netherlands. House prices dropped and owneroccupiers did not sell their house. The mobility rate dropped. To make a sound comparison through time, the focus is on two time periods, one before (2005)(2006)  This is the household type that is most associated with suburbanisation. In the Netherlands, when children are 5 years old, they have to attend school. After this transition, mobility rates generally drop steeply. This is for many families the moment they do not want to move, because otherwise children have to switch schools. By measuring the household situation at the end of the period (after a possible move), also moves just before the first child is born are included as well as moves where two partners to make the subgroups. Ideally educational attainment level should be included, because this is also an important driver in moving behaviour.
As this variable has too many missing values (about 25%), including this variable would lead to the undesired exclusion of too many families.
The first division is made by ethnic background: • one or both partners Dutch background; • both partners Surinamese, Antillean, Turkish and Moroccan background (first and second generations together); • both partners other migrant background or mixed from different migrant backgrounds (also both first and second generations).
The second division is by place of birth: • both born in the city; • one or both born in the municipalities in the region of that city; • both born elsewhere in the Netherlands or mixed born elsewhere in the Netherlands with born in the city and with born abroad; • both born abroad.
This division is only made when relevant and feasible not to end up with too small groups. The last division is by income level (middle and high income). The income level is based on the disposable household income, divided in quartiles based on the national distribution of income of all households. The high income group is formed by the fourth quartile, the middle income group by the second and third quartile. All young families with a low income (first quartile) form the last group. Low-income families have very little options on the housing market, and this will not differ that much within this group. Therefore, they are put in a separate group. Young families are quite an affluent group, mostly because many contain two working adults (74%). Approximately 50% fall in the fourth quartile. Only 5% has a low income (first quartile).
This results in the following 13 groups (see also  13. Low income.
The first two groups represent the 'original city dwellers families'.
Especially those with a middle income (Group 1) are historically associated with suburbanisation. Groups 3 to 6 represent the 'gentrifier families'. Groups 3 and 4 grew up in the urban region, which probably makes them more likely to return to the region around the city than Groups 1 and 2. Groups 5 and 6 grew up in the Netherlands outside of the urban regions and often came to the cities for education. Especially, Group 6 is mainly higher educated (over 70%). Groups 5 and     The final category of low-income families (13) (Blinder, 1973;Oaxaca, 1973). This technique helps to determine The Oaxaca-Blinder regression decomposition technique allows for distinguishing between both types of effects and quantifies the impact of both elements (Blinder, 1973;Oaxaca, 1973). Shuttleworth, Cooke, and Champion (2018) and Cooke (2011) used the technique recently for a similar issue. In this article, we follow their operationalisation: indicates the purchasing value at that moment. In the analysis, the situation before and after the move is measured in a combined variable.
Other control variables are duration-of-stay prior to the research period, whether the head of the household is a student, the educational level and type of education, the age of the head of the household and the number of children in the household. Table 1 gives the bivariate analysis of the 13 subgroups of young families by the control variables used in the regression analysis.

| RESULTS
The total observed population of young families contains 114,762 households. Ten percent of these young families moved to the region around the city in the two observed time periods (2005-2006 and 2015-2016). These outmigration rates differ between the 13 subgroups (see Table 1). In Groups 3 and 4, the 'gentrifier families' that originate from the region, indeed, have the highest outmigration rate (15% and 17%, respectively). They are followed by Groups 5 and 6, the 'gentrifier families' that originate from outside the region around the cities (10% and 12%, respectively). The outmigration rate of 'ethnic minority families' is very low (Group 7: 4%; Group 8: 5%).
In all the four cities, the population of young families has grown between 2006 and 2016. Families have become more numerous in cities. The increase of families outpaced the total growth of number of households (CBS, 2019). This could indicate that the preference for families has become more urban. But at the same time, the group of young families that moved from the city to the region of that city has also grown. In particular, in Amsterdam and Utrecht, the outmigration increased (+395 young families in Amsterdam and +303 in Utrecht, see Table 2).
The composition of the young families in these four cities differ from each other (Figure 1). In Amsterdam and Utrecht, higher-income families constitute the largest group. Most of them are of Dutch background and especially those that were born elsewhere in the Netherlands (Group 6) form a large group. This is also the group that has increased strongest in the 10-year time period. Both cities have large universities that attract many students from all over the country (and from abroad). It seems that indeed, a substantial share of the middle classes stay and form a family in the city, often referred to as 'gentri- In Amsterdam, the increasing numbers of suburbanizing families also originate from the growing group of 'expat families' (Group 12).
Again, this seems largely a result of composition, not of a change in moving behaviour.
In The Hague and Rotterdam the growth in outmigration of young families was smaller, and also not strongly related to any specific subgroup. In The Hague, the increased outmigration for the young families of Dutch descent was largely a rate effect. The outmigration of young families born abroad was a compositional effect. In Rotterdam, the moving behaviour of young families was quite stable, both the rate and compositional effects are small.
In total, the amount of young families moving from the cities to their respective surrounding regions increased by +1,108. All the compositional effects added together (per group and per city, all the black bars in Figure 2) reveal that +706 of this increase can be ascribed to the changing composition. The rate effects added together leads to +294 (grey bars in Figure 2 added together); these can be ascribed to a change in the rate effects. The remaining residuals add up to +108 (black rimmed bars). This means that 706 of the total increase of 1,108 (64%) can be explained by the growth of the amount of families living in the city. The group of young families in the cities has grown and therefore the outflow to the region around the cities has grown.
On top of that, the growth was mainly found in the 'gentrifier families' with a high income (Group 6), a group with a relatively high outmigration rate (12%). The majority of the increase of families moving to the region is therefore caused by a change in the composition of young families in cities.

| Moving behaviour of young families differ by residential biography (place of birth, migration background) and income
The Oaxaca-Blinder technique reveals which part of the absolute changes in numbers is compositional and which part is a rate effect, can be determined more closely (Table 3).
This analysis is split in two: The first set of models looks at the moving behaviour compared with not moving; The second set of models focusses on the movers alone. Those young families that originate in the region around the city have the highest odds for moving from the city back to that region (Groups 3 and 4). This is almost regardless of income level. This result is in line with previous research on residential biographies that people in the family formation phase often return to where they grew up (Blaauboer, 2011;Feijten et al., 2008).
The young families that are historically often associated with suburbanisation, the 'original city dwellers' (Groups 1 and 2, one or both Dutch background, both born in the city), are indeed a group that shows relatively high odds for moving to the region around the city.
Although waning in numbers, in their moving behaviour, they are still a relevant group in the suburbanisation process. The 'gentrifier families' (Group 6, one or both Dutch background, born elsewhere outside the region, high income) have slightly higher odds for moving to the region than the 'original city dwellers' (Groups 1 and 2). Although region. This is even lower than the reference category (Group 13, low-income families). But also when their income is higher, the odds for this group for moving to the region are relatively low. Table 3

Model 2 in
for moving to the region than one-or no-earner families. The highly educated have higher odds for moving, especially from the city to the region around the city. Young families that are educated in the social sciences less often move from the city to region than young families with education in other disciplines. This finding is in line with results from other studies that showed that people with more cultural capital are more urban oriented (Boterman et al., 2017).
The older young families are the lower the odds that they will move (both within the city as to the region). The larger the family (number of children), the higher the odds for moving to the region and the lower the odds for moving within the city. Young families with a short duration-of-stay on the current address have lower odds for moving (both within the city as to the region).
The model fits in this analysis of movers versus nonmovers is modest (Nagelkerke R 2 of 0.071 in the second model). The models give insight in the moving behaviour of young families with different socio-economic and socio-cultural characteristics, but only characteristics of the household are not sufficient to describe the moving behaviour. Especially, the characteristics of the housing market could be an important factor in explaining the moving behaviour of young families. These are added in the following analysis.

| Differences in the local housing market lead to different rates in outmigration
The second set of models starts with the same analysis, but in a binary logistic regression (Table 4). The first model is the same as Model 2 in Table 3 (Table 4). This indicates that the local situation on the housing market has a different outcome for young families in Amsterdam than for young families in the other cities. The third model therefore looks at interaction effects between the city and the housing characteristics. The average house size in Amsterdam is much smaller than in the other three cities, and this results in a high outflow of young families in Amsterdam as evidenced by the effect of moving to a larger home. In the other three cities, this is less the case. In The Hague and Rotterdam, moving from the city to region of that city is often related to moving to a more expensive home. But in Amsterdam, and to a lesser extent Utrecht, this seems not to be the case. When young families in Amsterdam move to the region, they more often move to a less expensive home. This seems to reflect the high house prices of Amsterdam, which are on average higher than in the municipalities in the region around Amsterdam.
Also, the year variable is lower in Model 3 (Table 4). This suggests that the increased outmigration of young families in 2015-2016 is partly related to changes in the housing market.
The odds for a few household types change when the housing characteristics are added in the analysis. The 'ethnic minority families' with a high income (Group 8) now have similar odds as the 'ethnic minority families' with a middle income (Group 7). Also, the odds for moving to the region for the 'expat families' (Group 12) are in the last model very low. This suggests that their slightly higher outmigration to the region around the city is related to housing characteristics.

| DISCUSSION AND CONCLUSION
Many cities grow demographically, largely caused by a continued inflow of young people from all of the country and also increasingly internationally (Beaverstock, 2012;Turok & Mykhnenko, 2007). These growing numbers of young urbanites boost the formation of new families in the city. As several studies claimed: gentrifiers increasingly settle down in cities, as families (Boterman, 2012). Yet especially highincome families continue to suburbanise. This paper investigated this complex issue by disentangling compositional and rate effects in suburbanisation processes in the four big cities of the Netherlands. Paradoxically, the odds that gentrifiers move to the suburbs in the region around cities are high compared with other types of young families, resulting in large numbers of families moving from the city to the region around the city. The increasing number of 'gentrifier families' in the city seems to be primarily a demographic process and not a result of changing residential preferences of young families. This stability in residential preferences is also found in other studies (Booi & Boterman, 2019;Fuguitt & Brown, 1990) The analysis suggests that family gentrification in cities can occur simultaneously with a growing outmigration of (then former) 'gentrifier families', possibly leading to 'spill-over gentrification' in other parts of the region (cf. Bridge, 2003;Paccoud & Mace, 2018).
The cities Rotterdam and The Hague were more stable in their composition of young families. The population of young families did grow, but not as fast as in Amsterdam or Utrecht. The young families in Rotterdam and The Hague are more often of middle income and/or ethnic minority groups, and the growth of the young families in these cities can be mainly found in this category. The young 'ethnic minority families', seem most attached to the city. Just few of them move to the region around the city. Although higher-income families within this group do show higher odds of leaving, it is still lower than the average. Suburbanisation is not absent within this group and some do move into home-ownership (Hamnett & Butler, 2010), but these migrant groups still show a strong urban preference and do not appear to follow the suburbanisation trend in the same pace as the original-Dutch population.
Our analysis shows that the intraregional moving dynamics are largely related to changes in the composition of the population. The increase in outmigration by about +1,100 young families between the two time periods (2005-2006 and 2015-2016) could for +700 families be ascribed to the changing composition of young families.
This was mainly caused by the growth of young families in cities in general and in particular the growth of gentrifier families with a high income (which have a high outmigration rate). The changes in moving behaviour (rate effect) appear to be smaller (about +300). This provides evidence supporting the urbanisation-literature arguing that changes in moving behaviour are mostly compositional (Buzar et al., 2005). The change in the rate effect, the increased chance that a family leaves the city to live in the region around it and the differences between the cities seem to be related to the changes and differences in the housing market. The most marked difference between Amsterdam and the other three cities is the influence of the size of the dwelling has on the outflow to the municipalities in the region around the city. The lack of larger homes suitable for families drives young families to the suburbs.
The families that are formed in gentrifying cities are increasingly of higher socio-economic status and less of lower socio-economic status. They have the possibility to choose on the housing market. They often tend to move to the region around cities to find better housing.
The case of Amsterdam shows that especially the size of the house matters. When the city does not provide adequate housing for this group, they will leave.
Although the Dutch urban dynamics contexts may have some peculiarities, this finding could also have relevance for other gentrifying cities. Especially, the trend, under high market pressure, to build small apartments could become disadvantageously for keeping families in the city.
In this study the main focus is on young families, because they are the main group that moves from the city to the suburbs in the region around the city. Notwithstanding the relevance of families for suburbanisation, by leaving out other household types we also excluded couples that may have moved to the region in anticipation of becoming a young family. Also young people who could not enter the housing market in the city and therefore moved directly to the region around the city are not included in the analysis. Both omissions may lead to an underestimation of the outflow from the city to the suburb related to family formation. Finally, the analysis here reduced suburbanisation to a dichotomy (staying within the city or moving to the municipalities in the region of the city). While providing a clear-cut insight into the moving behaviour of young families of different composition, metropolitan regions contain a plethora of rural, suburban milieus and even more urban milieus. Future research might investigate this diversity of suburban milieus. A more detailed and in depth spatial analysis of residential orientations of young families may be required before we will reach sufficient understanding of their moving behaviour.