Abstract
Finland has been known for its excellent PISA results in educational outcomes throughout the last decade. The country has boasted a rare combination of high overall level, as well as uniquely good outcomes of the bottom performers. However, the latest PISA results and the recent socio-spatial developments within the Finnish cities challenge this nationally celebrated balance in schools and urban social structure. Until now, research evidence has demonstrated that in the Finnish context with a powerful, universalist welfare state and a highly educated, homogenous population, differentiation increases mainly by the growth of an elite. Our analysis of large datasets from schools and neighbourhoods in Helsinki suggests that this development has been overturned in the local level: segregation has begun to increase and appears to operate through the trends of middle-class avoidance and the decline of the underprivileged groups in urban schools and neighbourhoods.
Introduction: The egalitarian PISA performer challenged
Finland has stood out as one of the highest performing countries in the OECD PISA assessments of educational outcomes throughout the last decade. The country has attracted global attention as the Nordic ‘PISA wonder’, which has been marked by an unusual combination of a large share of high performing students, as well as a particularly small share of poor performers in its comprehensive schools. At the same time, the country has continuously ranked extremely high globally in terms of income equality and other indicators of the social and spatial distribution of welfare (Anttonen et al., 2012).
The peculiar nature of the Finnish distribution of welfare and educational success is illustrated by a comparison of student percentiles in the PISA assessment of Mathematics 2003 (Figure 1). The graph represents all the students ranked from the lowest performers to the highest, and their score points are compared to the OECD average for the corresponding group (OECD average = 0). While the best performing Finnish students are around 20 score points above their international peers, the poorest performing quarter of students are approximately 70 score points ahead of the corresponding group elsewhere. The latter difference equals almost two academic years of learning. This pattern is mirrored in the schools, which have been among the least segregated in the world and between which the variance in educational outcomes has been very low (Kupari et al., 2013; OECD, 2010).
Differences in educational outcomes in student percentiles across countries in PISA 2003. The graph describes how students in different outcome categories compare with their counterparts in other PISA countries. The OECD average for each percentile is 0 (Törnroos and Kupari, 2005: 25).
The egalitarian trend observed in education has been one of the key features of Finnish society. A number of studies have described a similar, internationally rare pattern of distribution within other aspects of the society, even in the urban residential patterns of socio-economic and ethnic groups (e.g. Vaattovaara, 2009; Vaattovaara et al., 2011). The variance in outcomes throughout society is relatively small, and socio-economic differentiation has mainly operated through the rise of the most privileged groups or the top deciles – not through the diminishing welfare of the less privileged. The long-standing understanding of Finnish society has been that the vast majority of people, as well as schools and neighbourhoods, are faring well and that any growth of inequality is mainly driven by the rise of the most privileged (Hiilamo et al., 2010; Uusitalo, 1999; Vaattovaara and Kortteinen, 2003; Vaattovaara et al., 2011).
The social trajectories of the Finnish welfare model have been strongly built upon educational institutions (Castells and Himanen, 2002; Esping-Andersen, 1999; Kautto, 2001). Compared with British, French or German educational systems, the Finnish system has never been as selective; on the contrary, opportunities have been provided to all social classes throughout Finland (Mäkelä, 1999). The national and local policies reflect these experiences and ideals of an egalitarian society. While there has been a continuous consensus on taking care of the small disadvantaged group, the national concern regarding questions of the growth of undesirable polarisation or segregation has typically been directed to the rise of elites. As an example of this paradigm, the planning documents of a new waterfront development in Helsinki in the 1990s state that, ‘While aiming at a positive image [for the neighbourhood] we need to make sure to avoid an elitist stigma’ (Helsinki City Planning Office, 1994: 43). In these policies, the primary mechanism for growing segregation has been interpreted to be the growth or voluntary segregation of the top strata of the society.
Declining equality
The latest international PISA assessments showed a turn in the historical trajectories: a marked and uneven decline in the country’s results. Not only did the score point average drop by an equivalent of half a school year (25 points) from the 2003 results quoted above, but the bottom 10% of Finnish schools fell markedly further than most schools (Kupari et al., 2013). In fact, the study pointed out a group of schools which even fell below the OECD average. The comparatively steeper decline in the ‘tail end’ not only targeted schools, as the results of the poorest learners also fell more than the average decline in results. The share of ‘very poor’ performers – unable to meet the minimum criteria for managing the basic mathematical tasks needed in a contemporary society – rose from 7% to 12%. The effect of socio-economic background on individual outcomes was also shown to have grown significantly, as the effect of one unit increase in the SES background grew from 28 points to 33 points (Kupari et al., 2013).
The newest PISA results challenge the assumption of the distribution of welfare and the mechanisms of segregation within society. Other recent empirical findings do not counteract, but rather appear to enforce the change observed in education. Recent studies have shown that after decades of expanding welfare and narrowing gaps between population groups and urban neighbourhoods, the differences have for some years been growing more marked throughout society and urban space (Kortteinen and Vaattovaara, 2015; Vaattovaara et al., 2011; Vilkama, 2011).
This article builds on the conflict between the theoretical assumptions of the Finnish welfare state’s ability to protect the society from segregation through decline, and the empirical evidence of growing segregation and cumulative decline in Finnish schools and neighbourhoods. These developments challenge the core assumptions of the Finnish welfare state and underline the need for a new analysis of the drivers of segregation in the Nordic welfare context. In many European contexts, the growth of segregation within neighbourhoods and schools has been attributed in large part to middle-class choices, such as strategies to avoid less privileged groups (Allen, 2007; Ladd, 2003; Musset, 2012; Östh et al., 2013; Söderström and Uusitalo, 2010). Is this the case also in the Finnish context?
The aim of our paper is to examine the relationship between school choices and urban segregation in the egalitarian Finnish context, in search of the mechanisms contributing to the growing differences, even those seen within the internationally acclaimed PISA results. The research setting builds on the theoretical framework of the process of segregation operating through residential choices and local institutions. Thus, we especially address the question of whether school choices intensify or alleviate the level of segregation within schools and neighbourhoods.
The research questions focus on the relationship between the Finnish welfare regime and choice as a possible mechanism for segregation. Can we observe the internationally described tendencies towards rejecting neighbourhood schools in areas with a lower socio-economic status despite the former national strengths, and does choice lead to increased socio-economic and ethnic segregation in schools and neighbourhoods, as well as growing outcome gaps between schools? In short, does choice contribute to segregation in the egalitarian Finnish context – and if so, does the process operate at the top or bottom of the distribution?
Several scholars have addressed the role of welfare states in the contemporary political and theoretical debate on the causes and effects of residential segregation, arguing for its strong intervening and balancing role, protecting institutions as well as city regions from processes of decline and social exclusion (Kautto et al., 2001; Musterd and Ostendorf, 2004). Against this background, the Finnish case provides a unique opportunity to observe the processes of school choice and segregation in the field where the Finnish state has clearly been the best achiever in the world, the PISA.
Choice and segregation in the European perspective
Global policies on school choice are often based on the assumption that choice has the potential to alleviate school segregation (Forsey et al., 2008). It is often argued that choice can weaken the effects of urban segregation on school segregation through creating more opportunities for residents of underprivileged neighbourhoods. Another typical argument is that choice can counteract school segregation by forcing the weakest schools to improve in quality in their need to compete for students (see e.g. Ladd, 2003). However, the research evidence provides a more mixed view.
The issue of education and school choice has been addressed as a central and delicate question, especially regarding middle-class families in various urban contexts. Rather than simply reflecting a middle-class ethos for seeking an elite education, families have been demonstrated to opt for choice in order to avoid exposing their children to lower standards of education or the ‘wrong’ types of socialisation (see e.g. Boterman, 2013; Bunar, 2010; Maloutas, 2007; Raveaud and van Zanten, 2007; Schindler Rangvid, 2007). A number of European studies suggest that school choices are socio-economically selective and include a tendency of middle-class families to use choice as a strategy to avoid underprivileged groups. School choices can thus act as a driver of growing school segregation (Allen, 2007; Ladd, 2003; Musset, 2012; Östh et al., 2013; Skolverket, 2012; Söderström and Uusitalo, 2010). In an OECD meta-analysis of international research evidence, Musset (2012: 33, 35) summarises the effect as follows: Research shows that parents prefer schools with populations ethnically and socioeconomically similar to their own. […] [W]hile choice can be seen as a mechanism that levels the ‘playing field’ and provides the same opportunities for all, the evidence shows that it may not have the intended effects: better-off families and more educated parents are the ones who exercise choice, and that will enjoy access to a wider variety of schooling options. […] Therefore, the introduction of school choice mechanisms can lead to segregation across schools and to more disadvantages for those who are worse off.
Similar findings have arisen from some Nordic countries, as school choice has been assessed to have significantly reinforced segregation in Swedish and Danish schools. According to Schindler Rangvid (2007), ethnic segregation in the schools of Copenhagen has reached nearly the levels typical in the US, markedly exceeding the levels of residential segregation within the city. Swedish studies have demonstrated that school segregation has increased continuously due to the policy of free school choice introduced in the 1990s. The changes cannot be explained by the increased segregation of residential areas or by schools’ different resources (Östh et al., 2013; Söderström and Uusitalo, 2010). In their analysis of Stockholm, Söderström and Uusitalo (2010: 75) describe the increase in the socioeconomic and ethnic segregation of schools: [T]his reform, which was supposed to reverse the effects of residential segregation on school segregation, actually increased segregation along all other observable dimensions, particularly along ethnic and socio-economic lines.
Boterman (2013) has pointed out that ‘residential practices and school choice are tightly interwoven in a “geography of education”’ (see also Butler and Hamnett, 2007; Butler and Robson, 2003). The suburbanisation of white middle-class families both in the US as well as in European contexts has been interpreted as a common residential and school choice strategy (Boterman, 2013). School-related choices are thus strongly linked to neighbourhood segregation, maintaining and reinforcing the development of differences. Assessments of the quality, reputation and student composition of local schools affect education-motivated parents’ migration decisions and are thus also reflected in housing prices (Cheshire and Sheppard 2004).
Patterns of segregation and school choice in the Finnish context
Following several international models, Finnish school choice policies are based on the assumption of creating more opportunities for all residents, independent of their residential location (Seppänen, 2003). It is thus assumed that school choice on the comprehensive level, introduced to the system in the mid-1990s, does not increase socio-economic or ethnic sorting, but can, in fact, even decrease segregation between schools.
Following these ideas, Finnish educational authorities have treated school choice as a mechanism to encourage positive differentiation (Seppänen, 2003). The choices have been believed to be relatively independent of the socio-economic background of students and of the school catchment areas, and any possible sorting effect has been attributed to a ‘cream skimming’ effect – and a slight boosting effect on outcomes – in the most popular, or elite, schools.
This view draws from the interpretation of processes of social exclusion within a strong, Nordic welfare state. The effects of social segregation have been assumed to be small, and social institutions – such as schools – to be relatively well shielded from processes of growing segregation and exclusion.
The relatively low levels of socio-economic and ethnic segregation in Helsinki have been believed to be a factor further counteracting school segregation. The city has implemented a policy of mixing different social groups in the planning of new neighbourhoods since the 1970s along with strong housing policies aimed at combating segregation, and the fairly even distribution of welfare among the population has been also mirrored in the socio-spatial structure of neighbourhoods.
Prior to the 1990s, immigration in the country was virtually non-existent. Today, the share of residents with an immigrant background is still under 20% within the city, but the pattern of distribution within this group is very even compared to the majority of other European capitals. However, the majority of recent studies point at growing socio-spatial segregation as a permanent trend since the 1990s. Income differences between neighbourhoods have grown throughout the last 15 years, and certain neighbourhoods have become noticeable concentrations of urban disadvantage by national standards. In some neighbourhoods, the share of the immigrant population is close to 40%, exceeding 50% for school-aged children (Vilkama, 2011).
Recent research indicates that urban spatial differentiation is linked to the ‘geography of education’ also in the Finnish context. The schools’ learning outcomes have been demonstrated to vary according to the socio-economic and ethnic composition of the school catchment area (Bernelius, 2013). There is also an indication of sorting effects resulting from school choices, as highly educated families have been demonstrated to be over-represented in the group opting for school choice, and the most popular schools tend to be found in privileged locations in Finnish cities (Bernelius, 2013; Kosunen, 2014; Seppänen, 2006; Seppänen et al., 2012). Corresponding to international observations, especially middle-class families nowadays compete over highly valued school places more extensively than before in Finnish cities and their schools (Kosunen, 2014; Seppänen et al., 2012).
Residential location is tied to access to institutions, especially in Helsinki, which strongly affects the available choices for schools. The vast majority of Finnish schools are public institutions with a local catchment area, and private or specialised schools without a catchment area are rare. The school market consists thus almost entirely of public schools with strong spatial links to their catchment area population (Seppänen, 2003). Currently almost one-third of primary school students and nearly half of secondary school students in Helsinki attend a school outside their catchment area; in other words opting out of the school they would be assigned to based on their home address. However, places in schools outside one’s own catchment area are limited, and the only way to ensure a place in a specific school - or to ensure the ability to avoid one - is through residential choices.
Thus, the recent research findings suggest the possibility of interrelated processes of growing neighbourhood and school segregation in the Finnish context. However, so far the studies have not combined the factors, such as socio-economic and ethnic spatial segregation, school choices, learning outcomes and residential choices in a single, quantitative setting in order to analyse the interdependencies between these elements. The individual studies on learning outcomes and socio-spatial segregation (e.g. Bernelius, 2013) or residential choices and schools (e.g. Dhalmann et al., 2014) have also not been thematically focused on identifying the specific patterns or dynamics of the processes of segregation.
Research design
Our research aims at a two-fold goal in order to explore the possible patterns of segregation: to analyse the mechanisms and effects of school choices on public schools in Helsinki and to assess the importance of schools to the residential choices of families, that is, school choices resulting from residential choices. Integrating these questions into the same research design is particularly important, as school choices and residential choices affected by schools form the main mechanism for school segregation within the Finnish context, which is defined by a strong network of public schools. We will focus particularly on Helsinki, as the national capital is the most diverse urban environment in the country, thus offering a fitting opportunity to analyse the differences and the mechanisms of choice.
We will focus the analysis on the following subcategories to answer the research questions stated above: (1) Are school choices socially and ethnically selective? If so, are they driven by the elite looking for entrance in the highest performing schools (according to the national pattern of rising elites), or do they reflect the European trend of the middle classes avoiding the weakest schools? (2) Do school choices result in the differentiation of learning outcomes between schools? In other words, do the learning outcomes of students opting for choice differ systematically from the outcomes of students attending their local school? (3) What are the factors affecting the migration decisions of families with children? What is the particular role of schools among the push and pull factors affecting residential choice?
The research design is based on two large datasets: (a) a combination of national educational outcome assessments and register data from Statistics Finland; as well as (b) a survey on residential choices. The first dataset is used in the analysis of the determinants and effects of school choices: we analyse whether school choices are linked to the socioeconomic and ethnic structure of schools and areas in Finland. In addition, we will examine whether the choices produce any differentiation in learning outcomes between schools. The second dataset is applied to the question of the effects which schools and segregation have on residential choices in the neighbourhoods of Helsinki.
The first or school choice dataset combines data from several sources. The statistical data are drawn from registers covering the years 2010 and 2011 (national statistics authority, Statistics Finland and the City of Helsinki Urban Facts), and include basic socio-economic and language indicators, as well as information on housing type. These data have initially represented several geographical scales (from building-level data to a statistical neighbourhood), and we have combined them using GIS software to represent school catchment areas (83 primary schools and 45 lower secondary schools). The analysed materials also include school-specific data on students’ language groups within schools, as well as school choice, containing information on how many students travel to each school from other catchment areas and how many students living in a catchment area choose another school. These data include all catchment areas and public schools in Helsinki.
We have also acquired educational outcome data, compiled by the Finnish National Board of Education and the Helsinki University Centre for Educational Assessment. The measure for educational outcomes is the percentage of correct answers on standardised tests. These data include the long-term average for the years 2000–2008 of Finnish and mathematics outcomes, as well as learning outcomes and secondary school selections from 2005. At present, the school selection data are the only complete data available on Helsinki in which learning outcomes and school choices are combined at the individual level.
The survey complementing the register datasets was conducted in the Helsinki metropolitan area during the autumn of 2011. To be able to target the movers as well as stayers in different types of neighbourhoods, the data were collected by Statistics Finland. The survey includes residential choices, residential satisfaction and willingness to move among native households living in different types of residential areas. The survey is particularly focused on comparing the experiences of families with children who live or have recently moved away from areas of immigrant concentration to those of families who live or have moved away from other areas. The cooperation allows us to combine register data (e.g. level of income and education, number of children, dwelling stock and the population structure of residential areas) to the questionnaire data. The target group of the questionnaire consisted of 29- to 54-year-old native Finnish people living in the Helsinki metropolitan area who had lived in the same residential area in the metropolitan area in the years 2008 and 2009. Out of them, a sample of 3000 was selected, which was divided into four strata according to the persons’ migration status and type of residential area, sorted by the proportion of immigrants.
The research framework on the school choice analysis is based on correlation and regression analyses, which are used to examine the statistical link between the students’ school choices in Helsinki and the socio-economic and ethnic structure of the catchment areas and schools. In this article, school choices refer to actual school choices, that is, students’ attendance in either their local school or a school outside their own catchment area.
The evaluation of the effects of school choices is based on comparing two scenarios of educational outcomes in public schools in Helsinki, with a design comparable to the one used by Allen (2007) in Great Britain. In this analysis, students beginning the 7th form in lower-secondary schools are artificially allocated to their local school based on information about their primary school. In the first scenario, students are artificially reallocated to their nearest lower-secondary school, and the school outcome averages are calculated based on this reallocation scenario. In the second scenario, the school averages are based on the actual student composition in each school, or the realised scenario in the lower-secondary schools of the city. The learning outcomes of schools generated by the artificial reallocation are then compared to the actual outcomes to examine the net effect of school choice or how the outcomes have been affected by school choice (see also Östh et al., 2013).
Results
School choices and student sorting
Our analysis demonstrates that school choices appear to segregate schools in relation to student ability in the Finnish context. In this analysis, we compared the neighbourhood reallocation scenario, where students were artificially returned to their nearest school, to the actual situation in the schools of Helsinki. The educational outcomes used represent the points in a national assessment (range 0–100). If the Finnish assumptions of low or non-existent social and ability sorting in choices are accurate, the artificial reallocation scenario should show the same – or an even greater – level of differentiation between schools than the actual choice scenario. However, the statistical analysis of these two scenarios reveals that the exact opposite seems to hold for the lower-secondary schools of Helsinki: the reallocation scenario clearly produces a pattern of lower differentiation and smaller ability stratification between schools (Table 1). Thus, the levels of school segregation would in fact be significantly smaller if all students were reallocated to their own neighbourhood school, meaning that school choice indeed produces an independent effect for differentiation between schools.
The educational outcomes in lower-secondary schools of Helsinki based on two scenarios of school choice. The first scenario is the ‘no-choice’ scenario, where students are artificially reallocated to their nearest school, whereas the second scenario describes the actual outcomes after school choices were made by students.
The most striking observation considering the universalist context is that school choices appear to produce the most noticeable effects in both the highest and the poorest performing schools. While the variance between schools increases statistically significantly, the largest effects can be seen in the top and bottom deciles of schools. The difference between the single highest and lowest performing schools grows after choices by more than one standard deviation in the whole dataset. Thus, the schools attaining the best outcomes seem to profit from school choices, whereas especially the lowest decile loses heavily in the process. For these schools, school choices seem to represent an outward stream of motivated students moving to other schools.
Statistical comparison of the students opting for choice and the students attending their own neighbourhood schools confirms the observation. While the ‘choosers’ obtain an average test score of 54.8, the test score for the ‘stayers’ is 52.4. This difference represents over a third of the standard deviation between schools. In addition, two in three (65%) of the students opting for choice choose a school which has a higher average performance level than the student’s own neighbourhood school. The effect created by the choosers themselves has been taken into account in this comparison by using the reallocation scenario of students as the assessment of the schools’ performance level.
School choices also appear to have a strong link to the socio-economic and ethnic characteristics of both the catchment area and the school. When examined in detail, correlation analysis highlights the mechanisms with which the choices operate. Table 2 contains the relations of schools’ student composition, learning outcomes and catchment area population structure to students’ school choices. Preference in choices indicates the proportion of students coming from outside the catchment area of a school, in other words, the school’s popularity among outsiders. Rejection, on the other hand, describes the proportion of students going to schools outside their catchment area and opting out of their own local school.
The statistical relationship between socio-economic and ethnic segregation in the schools and school catchment areas, and the schools’ popularity or rejection in school choices in Helsinki.
Notes: Statistical significance * = p > 0.01, ** = p > 0.001.
Perhaps surprisingly, the rejection of schools is particularly consistent with socio-spatial segregation. This factor is strongly linked to school catchment area deprivation. A high percentage of immigrants (R = 0.40) and poorly educated adults (R = 0.30) are particularly statistically significantly linked to the school being rejected in school choices, whereas high income level (R = -0.32) and a high education level of adults (R = -0.35) are linked to a smaller likelihood of the school being rejected in the choices. The statistical importance of immigrants partly reflects the nature of segregation within Helsinki: the most multi-ethnic neighbourhoods are typically also socio-economically deprived areas, where particularly the ethnically Finnish population represents the lowest socio-economic strata.
In the case of primary schools, opting out of the local school is most strongly linked to the proportion of foreign language-speaking students in the school and its catchment area. On the other hand, the local school is the chosen school most often in areas where the proportion of highly educated residents is large and the yearly income high. These factors are even more strongly related to school choices than learning outcomes.
As for secondary schools, the relations are somewhat weaker, probably due to the smaller number of schools, but qualitative differences can also be observed. In secondary schools, increases in school learning outcomes and the catchment area’s yearly income are more clearly linked to choosing to attend the local school, whereas the relation between the proportion of foreign language speakers and rejection is somewhat less clear. In the case of secondary schools, there is also a statistically significant link between schools’ learning outcomes and their popularity in school choices, while in primary schools learning outcomes appear to have almost no relation to popularity.
The weak link between the schools’ popularity and neighbourhood characteristics is that the schools’ popularity is affected by a greater number of factors than school rejection. These may include, for example, the school’s reputation, the school’s overall ability to accept students from outside its own catchment area, as well as transport accessibility with respect to other schools. However, in the light of processes linked to segregation, the consistency of the relationship between the rejection of schools and the qualities of the neighbourhood is both a significant and a largely unexpected finding, pointing to processes contributing to segregation through decline in certain schools and neighbourhoods.
Schools and residential mobility
Our analysis of residential mobility further elaborates the link between choice and segregation. As explained earlier, the network of public schools with physical catchment areas ties school choices and residential choices together in an embedded geography of education. Even though our analysis of the motivators of out-migration shows that most residential choices are based on the quality and availability of housing (45%) or the price (21%), rather than educational facilities or other services, among the important reasons for the migration decision of families is the parents’ experience that a certain area was an unsuitable environment for raising children (12%).
The parents also assessed the significance of various factors for their decision to move on a four-level scale from very important to not at all important (Figure 2). Slightly more than one-fourth of all respondents regarded the unsuitability of an area for raising children as a very important or fairly important reason for moving away. The replies of the parents who had migrated away from immigrant-dense and socio-economically weaker areas differed significantly from those of the parents who had moved away from other areas. While 24% of the latter group regarded their area’s unsuitability as a children’s environment as an important reason for moving away, the number was nearly double (43%) in the first group. Thus, the qualities of the social environment emerged as the most important reason for moving away among those who had migrated away from immigrant-dense areas.
The proportion of respondents who considered the factor in question a ‘very important’ or ‘fairly important’ reason for moving away from their previous residential area.
In the Finnish context, schools do not appear to be a particularly powerful factor for the decision to move into a certain neighbourhood. Listing the four most important factors, only about 1% of the respondents mentioned the quality of the local school as the most important reason for choosing a certain residential area, and about 9% defined it as one of the four most important factors. Measured on a four-level scale (very much–not at all), one-third of the respondents estimated that the local school’s good quality had influenced their choice of residential area very much or fairly much. Other selection factors, related to a good environment, had affected their choice considerably more than the characteristics of the local school. It would thus seem that parents do not have to worry about the quality of the local school if they choose an environment whose social and physical characteristics correspond to their idea of a good child-raising environment – a good school forms part of the ‘package’ (see further Dhalmann et al., 2014).
On the other hand, schools appear to be push factors in certain types of environments (Figure 2). In socially deprived, multi-ethnic neighbourhoods, schools appeared to be more important motivators generating out-migration. In the city as a whole, only 5% expressed dissatisfaction with the local school. The results imply, however, that satisfaction with the local school diminishes with increasing segregation. All the parents who were very dissatisfied with the local school resided in immigrant-dense neighbourhoods. For the out-movers from these areas, the availability of a good school was three times more (40%) important than for the movers from other types of neighbourhoods (15%). Based on our targeted survey, a good learning environment is seen as part of the parents’ consideration of a good environment for children to grow up.
Our findings thus suggest that the issue of educational institutions and their student composition becomes increasingly important in segregated, multi-ethnic neighbourhoods, placing more importance on middle-class avoidance as a driver of segregation. Based on the questionnaire replies, parents’ ‘tolerance limit’ concerning the proportion of children with an immigrant background seems to lie somewhere between 20 and 50%. An analysis of the respondents’ wishes about the number of immigrant children in their child’s school in relation to their estimate of the current number confirms that the proportion of respondents wishing the number was lower is growing where the current number of immigrant children in the school is highest. When the estimated number of immigrant children rises to 20–50%, the proportion of parents happy with the current situation falls to about 30%, and nearly 45% wish that the number of immigrant-background students was smaller.
Determining the possible threshold values exactly on the basis of the figure, however, is not possible, because the classification given in the questionnaire is too rough for that purpose. Nonetheless, our questionnaire and interview data clearly indicate that considerable differences between residential areas and schools in the proportion of foreign language-speaking children affect the migration behaviour of families with children. Since, as known, parents also follow each other’s example in their actions, the fact that certain kinds of choices become more common and visible may render them more generally acceptable, even normative, and further contribute to the development.
Finally, the parental survey underlines the importance of school choice, thus connecting to our previous analysis on school choice regarding other catchment area schools. The factors related to the quality and student composition of schools are emphasised among the small group of parents who were dissatisfied with their local school. Out of them, nearly 70% had chosen to opt out of the local school. Half of these respondents mentioned the quality, reputation and ethnic student composition of the local school as reasons for selecting another school.
On the whole, our observations on urban schools suggest a continuous network of interactions, which together contribute to a process of segregation (Figure 3). Firstly, school segregation is affected by the socio-spatial segregation within the city: the structure of neighbourhoods forms the basis of the initial student composition and learning outcomes of schools. This, in turn, affects families’ school choices. As students who opt out of their local school achieve better than average learning outcomes, school choices seem to lead to a flow of well-achieving students moving away from rejected schools towards popular schools. This pattern of choice further reinforces the differences between schools. Finally, these developments may result in growing socio-spatial segregation in urban neighbourhoods, as the residential choices of families are clearly affected by their responses to local schools, especially in disadvantaged neighbourhoods.
The interrelated factors contributing to school and neighbourhood segregation.
Concluding discussion
Based on the analysis presented above, it is evident that processes of school and neighbourhood segregation operate in the spatially balanced Finnish context. We have shown how school choices have the effect of increasing the variance of educational outcomes between schools, and that both the school choices and the initial differences between schools are linked to urban segregation. The growth of urban socio-economic and ethnic segregation in the city is reflected in the population structure of the school catchment areas, which seems to act as a driver of even further differentiation of the schools’ student base and educational outcomes. This, in turn, appears to lead to growing pressure for school choice, as the rejection of neighbourhood schools is closely linked to the level of socio-economic deprivation in the catchment area.
As described earlier, the historical trajectories in education and residential developments have led to a widely held theoretical understanding of the Finnish welfare state and educational system as qualitatively different systems from more socially stratified societies, shielded from processes of segregation operating through cumulative decline.
However, our analysis demonstrates that the average learning outcomes of the bottom three deciles are clearly deteriorating as a result of school choices, while those of the top deciles are improving. Although possible differentiation between schools was believed to stem more from the existence of a few elite schools than from the relative decline of lower-achieving schools, our results indicate that population structure has the most consistent effect on determining school choices in the most disadvantaged neighbourhoods, where families opt out of the local schools.
Socio-spatial research has recently challenged the belief of a society shielded from processes of cumulative decline by pointing out that urban segregation appears to have begun to operate also – or possibly even mainly – through the decline of disadvantaged areas (City of Helsinki Urban Facts, 2013; Vaattovaara et al., 2011, Vilkama, 2011). According to the analysis presented in this paper, the evidence points to a similar conclusion in the case of schools. Segregation between school catchment areas and the schools’ educational outcomes is increasing, and the growth of differentiation is not only linked to the most advantaged areas and the schools gaining an extra increase in the process. On the contrary, urban differentiation and school choices are producing a noticeable effect on both ends of the statistical distribution. In disadvantaged neighbourhoods, the rejection of schools is systematically linked to the low educational and income level of the population.
International studies indicate that school-related choices are closely related to residential area segregation through the selective location choices of families with children. In light of this research, the quality of schools does not emerge as a clear single regional pull or push factor in the Helsinki metropolitan area. Rather, parents view the school environment as one part of the good child-raising environment they pursue through their choice of residential area. On the other hand, the results also imply that, as a result of the segregation development, the question of the choice of school is becoming more important to parents who make choices about where to live.
While the Finnish educational system is still a comparably equal one and produces excellent outcomes in international comparisons, the analysis points to its vulnerability to the segregation operating in neighbourhoods and institutions. Although the educational system and the welfare state are able to produce a comparable level of equality in the city of Helsinki, the observations challenge the assumption of specific qualitative differences in the processes of segregation within a system tied to the egalitarian context. Rather, the evidence appears to suggest that despite its apparent effectiveness in the former situation of very low levels of segregation, the welfare state and its educational system are, in the current situation, facing similar challenges as the majority of OECD countries.
The operational logic of urban and school segregation seems to remain constant in this context, with a marked difference only in the observed level of segregation, which is still internationally relatively low. If this interpretation is correct, it means that Finland, the global ‘model student’ in the PISA assessments and the country with ‘the worlds’ most equal schools’, appears to be facing major challenges and the threat of growing polarisation driven by social and urban change; an observation which contributes to the global understanding of the shared characteristics of segregation processes and of the role of choice as a potential driver of segregation.
Footnotes
Funding
This research received funding from the Academy of Finland and NORFACE Research Programme on Migration.