In recent decades, declining fertility rates, rising life expectancy and net immigration have reshaped the populations of European countries. These demographic developments are closely linked to both socio-economic and spatial processes. On the one hand, the profound changes in educational and occupational structures in post-industrial societies mean that younger birth cohorts are more likely to have completed tertiary education than previous cohorts (Eurostat, 2023a). On the other hand, these demographic and socio-economic changes have exacerbated urban-rural disparities. In particular, economic pathways have diverged between post-industrial metropolises and peripheral regions (Moretti, 2012). Rural communities and small towns have experienced population decline and ageing, whereas urban areas have grown significantly (Eurostat, 2023b). Economically strong regions are experiencing especially large population gains (Iammarino et al., 2019). In eastern Europe, population growth has often been confined to capital cities (Turgel and Ulyanova, 2023), while in France (Institut national de la statistique et des études économiques, 2019), Germany (Statistisches Bundesamt, 2024) and the United Kingdom (Office for National Statistics, 2021), population growth has occurred in the majority of large cities, albeit to varying degrees.

Rising spatial disparities are reflected not only in differences in population dynamics, but also in the changing socio-demographic composition of urban and rural populations (Eurostat, 2023b). In particular, the younger and more educated groups are seen as the drivers of urbanisation processes (Kashnitsky et al., 2020).

Another line of argument refers to increasing class differences in residential patterns. According to authors such as Florida (2004) and Reckwitz (2021), the representatives of the post-industrial economy, i.e. the “creative class” or the “new” middle class, are essentially urban classes who are attracted by vibrant and diverse metropolises. Moreover, the changing social composition of large cities is supposed to reinforce urban-rural disparities in population structures. Rising spatial inequalities are, in turn, expected to fuel social, cultural and political conflicts in many Western countries between metropolitan based cosmopolitans and provincially anchored communitarians (Zürn and de Wilde, 2016) – or between “Somewheres” and “Anywheres” (Goodhart, 2017).

Taken together, the literature suggests that in recent decades, “large-scale” socio-spatial disparities have been increasing. These processes not only fuel unequal regional development and urban-rural divides, but are also assumed to provide a structural basis for social and political polarisation and right-wing populism.

Following on from these observations, our study aims to investigate in more detail differences in place of residence between sub-populations defined by educational level in selected European countries. While the existing body of research has mostly analysed disparities between regions or districts – for example, regional differences in economic performance, infrastructure or population structures – our study applies a different analytical approach by examining how population groups are distributed over settlement types. ¹ By addressing spatial disparities between educational groups, our study intersects the perspectives of spatial demography, urban geography and social stratification.

Within this frame, we specifically ask whether socio-spatial inequalities have increased since the turn of the millennium. To answer our questions, we use data for six European countries from the European Social Survey (ESS) for the period 2002 to 2022. In the following, we first review studies that have dealt with issues of spatial disparities, and then derive our research hypotheses based on this review. After describing our database, we report the results. Finally, we discuss the relevance and implications of our findings.

Much of existing research on regional and urban–rural inequalities focuses on disparities between spatial entities (mostly regions, districts) and relies on indicators such as economic structure and performance, labour market features or demographic characteristics (e.g. Sabater et al., 2017; Gutiérrez Posada et al., 2018; Le Bras and Warnant, 2021; Brixy et al., 2022). Cross-national studies at the regional level have confirmed that disparate developments have occurred between capital regions and the rest of the country, which are reflected in employment and population structures as well as in economic performance (Eurofound and European Commission Joint Research Centre, 2019, p. 7). A large body of urban research has also dealt with socio-spatial inequalities at the level of cities. For instance, Musterd et al. (2017) analysed segregation patterns in 12 European capitals (Amsterdam, Athens, Budapest, London, Madrid, Oslo, Prague, Riga, Stockholm, Tallinn, Vienna and Vilnius), and concluded that socio-economic inequality has increased in European metropolises (see also van Ham et al., 2021 for a global perspective). Similar patterns of rising residential segregation have been observed in other European cities, including Brussels (Costa and de Valk, 2021), Berlin (Blokland and Vief, 2021) and Paris (McAvay and Verdugo, 2021).

Furthermore, demographic studies have shown that many European metropolitan regions have been experiencing population growth in the new millennium (Salvati et al., 2019, p. 22). Research on Germany (Sander, 2014), France (Labosse and Thouilleux, 2021), the UK (Thomas et al., 2015) and Spain (Lopez-Gay, 2014) shows that migration gains in large cities are driven by a continuous influx of young adults seeking education or work. These groups are mainly attracted to city centres, leading to talk of a “youthification” (Moos, 2015) of core cities (Siedentop et al., 2017). In contrast, families with children tend to move to the outskirts of major cities (Sander, 2014; Eder et al., 2018; Booi et al., 2020; Andersen et al., 2022).

While regional, urban and demographic studies have provided in-depth assessments of spatial disparities within and between cities and regions, researchers have paid much less attention to the issue of spatial disparities from a social stratification perspective. By this we mean population-wide studies that look at how socio-demographic groups are distributed across different settlement types – as opposed to research that focuses on spatial patterns on a sub-national and often local level, such as the studies described above. Recent studies on spatial disparities between social strata or classes include Konietzka and Martynovych (2022, 2023) for Germany and Vigna (2023) for France. While the latter found that average subjective social status is higher in large cities than in other settlement types (Vigna, 2023, p. 710), the former examined how socio-economic classes differ by place of residence. The results indicate that the higher classes are more likely to live in the metropolises, and that there was little change in this pattern over a two-decade period (1996–2018) (Konietzka and Martynovych, 2022, p. 194, 2023, p. 13).

When analysing the relationship between spatial and socio-economic characteristics, it is appropriate to distinguish between an aggregate and an individual-level perspective. Cities and regions may differ in their socio-demographic composition (social structures), and socio-demographic groups may be distributed differently across settlement types. These two phenomena are related, but represent separate analytical aspects of socio-spatial disparities.

In the context of our research question, it is important to note that changes in the demographic composition of settlement types are not necessarily accompanied by corresponding changes in the behaviour of individuals and groups. In other words, rising proportions of higher educated strata in large cities do not necessarily indicate that the higher educated have become more likely to live in large cities. The proportion of people with higher education may well have increased because the more recent birth cohorts are, on average, higher educated. At the same time, while it is plausible to assume that, at the individual level, the higher educated have a disproportionate preference for living in the metropolises, this does not necessarily mean that they are a numerically dominant population group in the metropolises.

Such considerations highlight the need to distinguish between phenomena that characterise settlement types or aggregate units (e.g. regions), and to address questions such as that of how cities differ in terms of their population structures on the one hand, and individual-level questions such as that of where different types of people live on the other. Against this background, our paper investigates in more detail how educational status and settlement type are related. We begin by examining changes in the population structure of different settlement types (metropolises, large cities, small towns and rural areas) across European countries (aggregate-level perspective). On this basis, we will look at how the residential behaviour of educational groups has changed in the new millennium (individual-level perspective).

In our analysis, we consider educational attainment as a proxy for resource endowment and social status in a more general sense, given that educational resources not only influence demographic behaviour (Mikolai et al., 2018; Zimmermann and Konietzka, 2018), but also have a fundamental impact on individual life chances (Baeriswyl et al., 2024).

Given the rising urban-rural disparities in European countries, the first thing we would expect to see is a growing divergence in the educational composition of the population in metropolises and small towns, and in urban and rural areas, over the last two decades. An extreme manifestation of this process would be polarisation between urban and settlement types, defined as a pattern in which the metropolises are increasingly dominated by higher educated groups, while small towns and rural communities are increasingly dominated by lower educated groups (hypothesis 1). Against this background, the question of to what extent individual-level dynamics have contributed to the increase in urban-rural disparities also arises. We thus examine whether the more privileged higher educated population groups are becoming more likely to reside in metropolises and large cities, while the less educated population groups are becoming more likely to live in potentially disadvantaged small towns and peripheral locations (hypothesis 2).

For our analysis, we use data from rounds 1 to 10 of the European Social Survey (ESS). The ESS is a representative scientific survey that has been conducted in European countries every two years since 2002. We include six countries (Belgium, France, Germany, Spain, Sweden and the United Kingdom) that have participated in all rounds since 2002, and for which consistent regional information is available over the observation period. The countries cover different geographical regions (northern, western, central and southern Europe) and socio-political regimes (universalist, liberal, continental and Mediterranean). In order to obtain robust case numbers for our analyses, we aggregate the survey years into three groups: 2002/07 (rounds 1 to 3), 2008/15 (rounds 4 to 7) and 2016/22 (rounds 8 to 10). ² We select respondents between the ages of 25 and 74, excluding younger age groups who are often still in education and whose residence may be temporary and determined by regional educational opportunities. We also exclude respondents with missing data, which gives us about 99,000 valid cases (for the analytical sample, see Table S.1, Supplementary material, available online at https://doi.org/10.1553/p-9g9e-pf8p).

Our dependent variable is the respondent’s place of residence. The dataset provides information on the self-reported place of residence, distinguishing between a farm or home in the countryside, a country village, town or small city, the suburbs or outskirts of a big city and a big city. On this basis we distinguish between rural communities (rural villages and farms or individual houses in the countryside), small cities (medium-sized and small cities) and large cities (big cities and the suburbs or outskirts of a big city). Furthermore, making use of the NUTS classification, we add the nations’ capitals as a separate category. ³ The data allow us to identify both Berlin and London within their administrative boundaries, while the Belgian capital of Brussels, the French capital of Paris, the Spanish capital of Madrid and the Swedish capital of Stockholm are merged with their surrounding areas. ⁴

Our core independent variable is the individual’s educational level. Based on the International Standard Classification of Education (ISCED), we distinguish between up to the lower secondary level (ISCED 0–2, “lower education”), the upper secondary or post-secondary non-tertiary level (ISCED 3–4, “medium education”) and the tertiary level of education (ISCED 5–8, “higher education”).

Gender (male, female), age group (25 to 34, 35 to 44, 45 to 54, 55 to 64, 65 to 74) and migration background are taken into account as control variables. Migrants are defined as respondents who were not born in the country of the survey (for descriptive statistics, see Table S.2, Supplementary material).

The article started from the idea that growing social inequalities in Western countries have also left their mark at the spatial level, not only in terms of residential segregation within cities, but also in terms of large-scale patterns of spatial disparities between urban and rural areas. As has been argued by social and political theorists (Florida, 2004; Moretti, 2012; Zürn and de Wilde, 2016; Goodhart, 2017; Reckwitz, 2021), the better-off are increasingly moving to the prosperous metropolitan areas, while the lower social classes are being pushed into rural or remote areas.

By looking at spatial disparities from a social stratification perspective, we have sought to add a population-based, individual-level analytical perspective to this discussion, which is often dominated by stylised juxtapositions of traditional and less educated rural population strata versus culturally diverse and higher educated urban population strata.

Using data from the European Social Surveys from 2002 to 2022, we examined the extent to which educational groups in six selected countries differ in their residential locations, and whether spatial disparities between these groups have increased in the new millennium. We used educational attainment as a proxy for socio-economic status and place of residence as an indicator of spatial disparities. Specifically, we tested hypotheses looking at (1) whether the composition of the population by education has diverged in large cities and rural areas, and (2) to what extent this divergence is due to differences in behavioural patterns between educational groups.

Regarding the differences in educational structures between large cities and rural communities/small towns, our analyses showed that, contrary to expectations, the social composition of both capitals/large cities and small towns/rural communities has developed in the same direction. We found that in all countries and settlement types, the proportion of higher educated groups has increased, while the proportion of lower educated groups has decreased. Although the dynamics of structural change vary from country to country, the overall picture is that the gap between settlement types did not widen over the period under consideration.

With regard to the second question on differences in residential location between educational groups, the data showed little change in residential disparities over time. Additional analyses that separated the younger and the older age groups did not substantially change the overall picture. Thus, our conclusion is that there have been no significant changes in the “large-scale” spatial concentration of educational status groups in the six countries since the turn of the millennium. At the same time, we found substantial country differences in the degree of spatial disparities. While a clear relationship between educational status and place of residence emerged in France and Sweden, the association was weak in Belgium and the United Kingdom, and was moderately strong in Germany and Spain. The reasons for these differences between countries are not obvious, as they do not follow regional or welfare state differences. Nevertheless, our analyses suggest that the socio-spatial sorting of population groups has not increased in any of the six countries. In sum, the findings run counter to the widely held belief that regional and urban-rural disparities have been steadily increasing in Western societies (Florida, 2004; Goodhart, 2017; Reckwitz, 2021).

However, even though the empirical evidence presented in this paper indicate that the discourse on spatial polarisation is exaggerating or generalising phenomena that may in fact be much more restricted in scope, we have to remain cautious about our conclusions. First, our analyses were limited to a time frame of just over 20 years, which may be too narrow to capture longer-term socio-spatial changes in the countries under consideration. It would be desirable to retrieve datasets that go back to the 1990s and 1980s. Moreover, it too early to tell if the Covid-19 pandemic has altered the attractiveness of large cities and reversed established migration patterns. Recent studies are reluctant to assess the medium- or long-term consequences of the pandemic for migration processes in European cities (cf. Rink et al., 2022; Stawarz et al., 2022; Wolff and Mykhnenko, 2023). Second, due to insufficient case numbers at the country level, we could only roughly classify cohorts or age groups. In particular, it would be desirable to investigate in more detail the residential choices of the younger highly educated strata. Third, we faced analytical limitations in terms of our classification of place of residence. Developing and applying more fine-grained and comparable classifications of city types (cf. Florida’s creative cities concept) are tasks for future research. Fourth, the ESS data imposed restrictions in terms of the spatial delineation of the capital cities and surrounding areas, which differed between countries. In a similar vein, our data did not allow us to map suburbanisation processes in large cities and metropolises. Finally, given that the educational structures in the countries under consideration differ considerably, using alternative socio-economic classifications may provide additional insights into the relationship between social stratification and place of residence.

Despite these restrictions, our analyses presented empirical evidence indicating that social-spatial disparities exist beyond the widely documented segregation patterns within cities and their surrounding areas, and, at the same time, that the degree of spatial disparities between population groups varies considerably at the national level across European countries.