The end of urban sprawl? Internal migration across the rural‐urban continuum in Switzerland, 1966−2018

Abstract In high‐income countries, migration redistributed populations from congested city centres into the sparsely populated outskirts, raising challenges to environmental and population health and the conservation of biodiversity. We evaluate whether this periurbanisation process came to a halt in Switzerland by expecting a decline in internal migration and a renewed residential attractiveness of urban agglomeration centres (i.e., re‐urbanisation)—two recent trend changes observed in Europe. Relying on data from censuses, registers and surveys, we describe trends in the intensity, geography and sociodemographic differentials of migration across consistently defined urban agglomeration density zones between 1966 and 2018. Although the overall intensity of migration declined, the rate increased among the working age population in part because of the societal diffusion of tertiary education. The dominant urban‐bound migration flows are increasingly confined within agglomerations over time. After the diffusion of periurbanisation down the city hierarchy between 1966 and 1990, we observe the emergence of re‐urbanisation in some agglomerations and sociodemographic groups around 2000. However, this phenomenon has been temporarily inflated by period‐specific transformations in Swiss society. More recently, the process of periurbanisation intensified again and expanded more and more beyond official agglomeration borders.

density built-up environments also increases energy consumption, may occur at the expense of forests, decimates invertebrate species, lowers ground-water reserves and affects population health through heat island effects. 1 In line with the United Nations' Sustainable Development Goal 11 ('make cities and human settlements inclusive, safe, resilient and sustainable'), 2 local authorities have issued various directives to limit urban sprawl. We question whether the process of periurbanisation came to a halt in Switzerland by investigating trends and patterns of internal migration across agglomeration density zones over the last 50 years.
This study addresses two important issues arising from recent international research that challenged established models of internal migration and urbanisation Salvati et al., 2019). We question whether the intensity of internal migration has declined, and whether its geographic pattern has switched from the centrifugal process of periurbanisation to a move back to central areas of agglomerations (i.e., re-urbanisation), leading to a renewed spatial concentration of urbanisation. The present case study on Switzerland -a highly urbanised and developed country in Europe-contributes to the existing evidence in two ways. To identify the spatial diffusion of urban migration patterns, we adopt a geographically detailed and internationally comparative perspective of the rural-urban continuum defined consistently over time and across space. The long-term perspective of this analysis, provides the keys to understand the importance of recent changes in migration and urbanisation in the context of the longstanding trends that continue to unfold.
After reviewing the recent evidence on internal migration and population redistribution, we present the data and methods. This is followed by a countrywide description of trends, sociodemographic differentials and the spatial focus of internal movements in Switzerland. We also map changing geographic patterns of migration in the 79 urban agglomerations and identify the population groups driving those trends. The paper concludes by a summary of the results and a discussion of their relevance for understanding the process of urbanisation in high income countries.

| BACKGROUND & RECENT EVIDENCE
The hypotheses of a mobility transition (Cooke et al., 2018;Skeldon, 2019;Zelinsky, 1971) and differential urbanisation (Champion, 2001;Geyer & Kontuly, 1993;Kontuly & Geyer, 2003) propose a patterned spatiotemporal diffusion and changing motives of migration during the transformation from a predominantly rural to an essentially urban and developed society. In early stages of this urbanisation process, the rising economic centre-the primate citypulls migrants looking for new income opportunities. With the regional diffusion of development, subnational income disparities attenuate, redirecting migrants towards lower-ranked cities. As a consequence, the city hierarchy diversifies. When the society is predominantly urban, rural-to-urban migration diminishes. Intercity and within-city flows are then expected to dominate migratory movements. We investigate whether the migration patterns of Switzerland correspond to this late stage of population redistribution.
The proportion of inhabitants living in urban areas indeed increased from 50% to more than 80% between 1960 and the early 2010s (BFS, 2017;Cunha & Both, 2004).
Massive migration into cities led to urban and industrial congestion effects in central areas. Consequently, jobs have been delocalised outward. People followed this move, thereby spatially extending the city borders [a process referred to as suburbanisation; (Champion, 2001)]. With the shift from an industrial to a postindustrial economy starting in the 1960s, a second phase of urban sprawl emerged due to multiple factors (Champion, 1989;OECD, 2018;Rubiera-Morollón & Garrido-Yserte, 2020;Shaw et al., 2020). Rising income per capita increased aspirations for more housing space (which is scarce and/or expensive in central areas of agglomerations), as well as for environmental amenities in less congested and more natural settings. The improvement of transport and communication technologies reduced the significance of distance to the workplace as a residential determinant. The diffusion of cars and large-scale investments in road and public transport infrastructure enabled people to commute over longer distances.
Thus, the second phase of urban sprawl extends into formerly rural areas of low population density, located on the more distant urban periphery; a process referred to as peri-or counterurbanisation (Champion, 1989). 3 This changing geography of migration over the stages of urbanisation-from centripetal flows into central areas of urban agglomerations to centrifugal movements towards the outskirts-has been confirmed by recent cross-sectional data in various world regions Rees et al., 2017;Rodriguez, 2007). Between the 1960s and the 1990s, Switzerland experienced intense periurbanisation of highly skilled individuals and upper-class families, leading to population decline in the centres of agglomerations (Cunha & Both, 2004). The present study traces the onset and spatial diffusion of this process over the last 50 years.
However, recent international research suggests a potential slowing down (and ultimately the end) of periurbanisation in highly urbanised settings. Firstly, the intensity of internal migration has surprisingly declined in many countries across the world since 1980although Europe was less affected by this phenomenon (Alvarez et al., 2021;Bell et al., 2018;Cooke, 2013). The drivers of this recent observation are not yet established. Kalemba et al. (2021) showed how population ageing determined the decline in migration within Australia over time. This trend has been only partly counterbalanced 1 Heat island effects refer to the stronger warming up of air masses on impervious surfaces (when compared to natural surfaces) due to limited reradiation of heat and evapotranspiration of trees, as well as increased traffic and energy consumption. 2 https://www.eda.admin.ch/sdg11-e 3 Urban sprawl manifests itself in different forms-such as through population distribution, the geography of population densities or land use patterns. In this paper, we focus on the underlying demographic behaviours, as constituted by internal migrations of the resident population.
by an increasing share of geographically more mobile groups in the population, such as international immigrants and highly skilled adults in working ages. Education indeed raises aspirations for mobility, increases its potential benefits and helps overcoming its financial barriers (DeHaas, 2010;Sjaastad, 1962). In Switzerland as elsewhere, individuals with more skills, better pay and higher professional status migrate more than the lower social strata Carnazzi-Weber & Golay, 2005;Charton & Wanner, 2001;Zufferey, 2020). Although Switzerland ranks among the countries with the highest intensities of internal migration in the world (Bell et al., 2015), we lack evidence on trends according to people's family stage and educational attainment at fine spatial scales.
A second unexpected finding in the literature relates to the renewed population growth in central areas of urban agglomerations since the turn of the 21th century in a number of European countries (such as in Italy, Germany, the Netherlands, Spain, the United Kingdom, etc.) (Dembski et al., 2021;Halbac-Cotoara-Zamfir et al., 2020;Kabisch & Haase, 2011;Salvati et al., 2019). 4 The drivers of this so-called re-urbanisation process are multiple, including market forces, the individuals' changing residential preferences and urban planning (Dembski et al., 2021;Rérat, 2012;Salvati et al., 2019). With economic globalisation, the rising economic sector of specialised services with high added value requires centrality to benefit from frequent interactions among agents and enhanced access to international transport and communication infrastructure.
To ensure international competitiveness, many cities have also regenerated central areas of agglomerations by constructing new business, cultural and residential districts. This has led to a gentrification process, which substitutes higher-class inhabitants to socially disadvantaged residents. Furthermore, an increasing share of the youth is attracted by central places in the context of the boom in higher-level education. Over the last decade, people also increasingly postponed the onset of family formation to later stages of their life course. As a consequence, adults can enjoy downtown city-life over a longer period of time before moving into the urban periphery to raise children. Finally, governments concerned with the environmental impacts of urban sprawl increasingly implement urban containment measures that aim at constraining the inhabitants' locational choices within existing built-up areas.
Following three decades of population decline, central areas of Swiss agglomerations experienced renewed population growth since the turn of the new Century. This was driven primarily by the arrival of an increasing number of international immigrants and, secondly, by the youth and the young and highly skilled working-age adults (Rérat, 2012(Rérat, , 2019. Switzerland's revised Spatial Planning Act of 2014 promotes this change in settlement patterns to densify cities and mitigate the environmental risks of urban sprawl. Many local and cantonal authorities have defined a moratorium on the zoning of new constructible land areas and issued directives to constrain future construction activities within the agglomeration extents (rather than beyond).
Assessments of urbanisation patterns from an land-use (rather than a demographic) perspective, by contrast, highlight a continuous expansion of settlement areas in Switzerland, although the pace slowed down recently (BFS, 2019). The proportion of the urban residents that lives in zones with less than 1.5 thousand inhabitants per square kilometre continues to decrease at an intensifying pace (OECD, 2018). Population re-urbanisation in Switzerland actually emerged alongside continuous demographic growth in the urban periphery, which remained the dominant form of urbanisation until recently (Rérat, 2012(Rérat, , 2019. The question is thus whether re-urbanisation is about to supersede periurbanisation. Previous research in Switzerland has emphasised international migration flows as the main driver of renewed population growth in central areas of agglomerations (Rérat, 2012(Rérat, , 2019. Agglomeration centres obviously constitute gateways for new immigrants (Cunha & Both, 2004), especially since the late 1990s when highly skilled individuals working in the knowledge-based sectors of the economy have dominated the inflows (Wanner & Steiner, 2018). However, this demographic contribution becomes increasingly temporary for two reasons. First, more than half of recent immigrants spent less than 6 years in Switzerland to develop their professional career before moving on to another country or returning back home (Fioretta & Wanner, 2017;Wanner, 2020). Second, those migrants, who integrated over a longer period of time in Swiss society, are characterised by higher levels of internal mobility than the natives (Charton & Wanner, 2001) and have adopted similar residential preferences for periurban zones (Lerch & Wanner, 2010). We argue that sustainable re-urbanisation over the long term depends on the redistribution of the resident population through internal movements, rather than on international migration flows.
Yet we lack evidence for a renewed shift in the geography of internal migration-from a centrifugal pattern (leading to periurbanisation) back to a centripetal one (driving re-urbanisation). It remains unclear whether re-urbanisation is consolidating or whether the urban containment measures only had a limited impact. In this context, we evaluate whether a potential decline and a change in the geography of internal migration has led to the end of periurbanisation in Switzerland.

| METHODOLOGY
We describe trends, sociodemographic differentials and spatial This so-called re-urbanisation process does not exclusively refer to population growth after a period of decline in central zones of agglomeration. Alternative perspectives in the literature include a final stage in the spatial cycle of urban growth (in which central areas grow more than the urban agglomeration belts), or the qualitative diversification of urban society [see Rérat (2012) for a theoretical discussion]. As stated above, we focus on the intensity and changing geography of internal migration that underly these multidimensional transformations. LERCH | 3 of 14

| Data and definitions
We rely on individual-level data from the full censuses 1970, 1980, 1990 and 2000, and the population register 2015−2018, which all provide information on migration status, and the municipalities of origin and destination. The censuses also include information on educational attainment. As this information is not available in the population register, we further mobilise the pooled annual waves 2015−2018 of the structural survey (which is administered to a 3%-sample of the registered population). The censuses provide the individuals' economic or de facto place of residence (i.e., weekly residents are enumerated at their place of work or study, rather than where they are registered). The register and the survey, by contrast, provide information on the legal residence. The discrepancy between the two approaches may be particularly important for young people.
The concept of usual residence is ambiguous at that age, especially for students spending the workweek in the place of higher-level education, while remaining registered in their parents' household. We are unable to adjust the census data to the legal definition in the register/survey data (because legal place of the migrants' origin is unknown). The uncertainty of the quality of information on usual place of residence calls for caution in the interpretation of the results-especially for the youth. 5 Based on the comparison of an individual's place of residence at the time of data collection with the self-declared place of residence 5 years earlier (i.e., migrant transitions), we analyse internal migration matrices for the 5-year periods 1966−1970, 1976−1980, 1986−1990, 1996−2000 In this analysis, we do not account for international migration flows for the reasons stated earlier and because our data only allow to observe international inflows, not the outflows. We focus on internal migration of the resident population as enumerated or registered at the dates of the respective data collections. Recent international migrants who arrived in the immediately preceding 5 years are excluded, but longer-term immigrants are included in the resident population.
During the observation period, Switzerland experienced sustained population growth from 5.8 million inhabitants in the mid-1960s to 8.6 million in 2019, as well as strong economic growth [the gross domestic product per capita (GDPc) rose from 13,000 to 84,000 CHF at current prices]. However, the trend in economic development was discontinuous, with periods of sustained growth and intermittent phases of stagnation and crisis. Our 5-year periods of observations can nevertheless be compared with each other, as they were are all marked by a positive economic context. Even in the last period observed (the aftermath of the global financial crisis), GDPc continued to grow-although at a significantly reduced pace.

| Spatial classification
To analyse migration across the rural-urban continuum, we regroup We further disaggregate urban zones in accounting for the ruralurban continuum of built-up areas. We rely on the SFSO's operationalization of Eurostat's new 'degree of urban' classification based on 300 x 300 m grid cells of the Swiss territory, aggregated at the municipality level (BFS, 2017). This internationally comparative definition of the rural-urban continuum has been adopted by the international community to promote comparative statistics on urbanisation worldwide. We disaggregate agglomeration zones in three population-density classes, that are organised in the form of concentric rings around the central municipality: densely populated areas or core cities (i.e., central municipalities with more than 500 inhabitants per km 2 ), semidensely populated places (corresponding to suburbs with 200−500 inhabitants per km 2 ), as well as the sparsely populated areas within agglomerations (corresponding to periurban zones) and those located outside of the agglomerations (see Figure 1).
If periurbanisation will not come to a halt, agglomerations will further extend into rural areas. We therefore defined an additional agglomeration zone, as constituted by a concentric ring of officially designated rural municipalities that directly border the agglomerations. Accounting for these extended agglomeration borders provides an interesting outlook on the potential future urban territory of Switzerland. In the case of continuous periurbanisation, the Swiss plateau on the North of the Alps will constitute one single and continuously urbanised mega-urban area, rather than a set of spatially independent agglomerations as observed in 2018. This scenario is at odds with the largely accepted urbanisation scenario of 'metropolisation'. The latter territorial vision aims to ensure the Swiss cities' international competitiveness in a globalised economy, while limiting the negative environmental consequences of sprawl. This is ensured by a set of distinct metropolitan regions-each of them 5 There is virtually no information on the quality of information on usual place of residence in censuses. Uncertainty may be high, especially for data on the youth. In the municipality of Zurich, for example, the net migration rate of the population aged 20−24 years increased sharply from 2.82% in 1986−1990 to 21.12% in 1996−2000. However, rather than decreasing in 2010−2018 (because weekly residence is not considered in the register data), the rate further increased to 29.14% in 2010−2018, even though the percentage of weekly residents dropped from 28% to 24% between 2000 and 2018 (according to data from Statistical Office of the municipality of Zurich). At the censuses 1990 and 2000, many young people may have indicated as their economic residence the place in which they were registered and have grown up, even though they spend parts of the week elsewhere.
including a large polycephal urban area centred around the most dynamic city and containing its inhabitants within a densely populated built environment (Cunha & Both, 2004;Kaufmann, 2012).
The analysis below refers to the extended urban agglomeration extents, thereby adding an additional differentiation of the urban-rural continuum of space. Hence internal migration is defined as a change in place of residence across 199 spatial zones, including the concentric agglomeration density zones (dense, semidense and sparse areas) and the ring areas of the 79 distinct agglomerations, as well as the remaining rural territory of Switzerland. This spatial disaggregation of the Swiss territory as of 2018 is applied to the data for earlier years. In other words, we control for the rural-urban reclassification of municipalities over the process of urban sprawl since 1966 to focus on migratory dynamics across consistently defined areas over time. Consistency in the spatial disaggregation was ensured by harmonising the municipality structure of the Swiss territory over the five decades under study in collaboration with the Statistical Office of the canton of Zurich, taking into account the administrative regrouping and/or splitting of municipalities, as well as the exchanges of territory between themas reported by SFSO in its historical registry of municipalities (BFS, 2021).
In the analysis of sociodemographic differentials in migration, we distinguish the following population age groups: children aged less than 15, the youth and young adults aged between 15 and 24, the

| Trends in internal migration
To investigate whether the intensity of internal migration has declined in Switzerland, Figure 2 shows the trend in the crude internal migration rate across the urban hierarchy and the agglomeration-specific density zones. The 5-year rate of migration of the total population indeed continuously declined between 1966 and 2018, but only slowly from 14.63% to 12.91%. F I G U R E 2 Crude rate of migration (over 5-year periods) across the city hierarchy and the rural-urban continuum, by age group, Switzerland 1966.Sources: Population Censuses 1970, 1980, 1990

| The Swiss-wide urban geography of migration
Switzerland's advanced stage of urbanisation conjectures major interand intra-agglomeration movements versus a marginal rural exodus.
Urban-bound migrants indeed clearly dominate, with a rather stable 82%−84% of migrants over time (not shown). Migrants moving between distinct agglomerations are the largest group, but their share is decreasing (from 58.11% in 1966−1970 to 47.69% in the last decade). Migrants increasingly move within agglomerations, especially between the late 1960s and 1970s (when the share increased from 23.54% to 29.75%, before reaching 33.97% in the last decade).
While the minority group of rural-to-urban migrants declined over time (from 11.37% to 9.21%), the percentage of migrants moving in the opposite direction increased (6.97%−9.11%). This points to a spatial extension of periurbanisation in Switzerland.
To evaluate whether the geography of migration shifted from a centrifugal pattern back to a centripetal pattern, Figure 4 shows the Alongside their increasing importance, their spatial pattern became more (rather than less) centrifugal over time. Among the out-migrants from dense areas, the percentage moving towards sparse and ring areas rose from 15.73% to 21.73%. The share of out-migrants from semidense areas settling in more sparsely populated agglomeration zones increased from 39.62% to 49.21%. Within-agglomeration migrants from sparse areas also increasingly focus towards the agglomeration ring (i.e., from 8.78% to 11.15% over the observation period).
The right-hand panel of Figure 4 focuses on migrants between agglomerations. A centrifugal pattern emerged over time, too. Among the out-migrants from dense urban zones, the share moving either to sparse and ring areas or into the rural territory increased from 20.05% to 26.23%. Among the out-migrants from semidense zones, the proportion heading towards sparse, ring or rural areas increased from 24.26% to 36.16%. Migrants from sparse areas also increasingly move to similarly sparse, ring or rural zones over time. Even the rural out-migrants more and more settle in sparse or ring zones (with increasing shares from 23.77% to 35.34%), rather than in the dense areas at destination.
In sum, the dominant intra-and inter-agglomeration migrants, as well as the marginal rural exodus, increasingly favour less densely populated zones in the periphery of agglomerations. This corresponds to a typical pattern of periurbanisation, rather than reurbanisation.

| Detailed spatial patterns of net migration over time
To analyse the spatial diffusion of periurbanisation and the emergence and potential consolidation of re-urbanisation across the 79 agglomerations, Figure 5 maps the 5-year net migration rates of the agglomeration-specific urban density zones and ring areas, as well as of the remaining rural territory of Switzerland for the five observation periods.
In the late 1960s, the rural exodus was still important. We used Kitagawa's (1955) method of decomposition of rate changes. The 5-year crude rate of net migration in a given agglomeration zone is a weighted average of the migration rates in different subgroups, with the weights being the groups' relative representations in the total population of that zone. In other words, changes in crude net migration in a given zone result as much from changes in group-specific migration rates (e.g., behavioural effects) as from variations in the relative representation of more versus less F I G U R E 4 Relative distribution of the inter-and intra-agglomeration migrants' destination zones, by origin zone, 1966−1970 and 2010−2018 (average 5-year period), Switzerland. Sources: Population Censuses 1970, 1980, 1990, Population Register 2015 F I G U R E 5 Net migration rates (over 5-year periods) in the agglomeration density zones and ring areas, as well as in the remaining rural territory of Switzerland, 1966−2018. The official agglomeration borders (excluding the agglomeration ring areas) are drawn in white colour; white coloured polygons represent lakes. Sources: Population Censuses 1970, 1980, 1990, Population Register 2015 migratory subgroups in the population (e.g., changes in the population composition).
We distinguished nine sociodemographic groups, as defined by the age classes introduced earlier and the differentiation by educational attainment of the young and mature working age population. We expect that the population aged 15−24 and the young working age population move into dense urban zones for educational and career development purposes. The mature working age population, by contrast, may look for places to raise children in socially desirable and affordable environments, situated in the less congested urban periphery. These sparse and ring areas may also be attractive to retirees looking for natural amenities, lower housing costs and well-developed urban (transport) infrastructure. Figure 6 only includes the behavioural contributions to the crude net migration changes over time (e.g., the effect of changes in group-specific rates, weighted by the groups' relative representation in the population), because these made up at least three quarters of the total contributions (contributions that represent less than 3% of the total variation are not shown to increase legibility of the figure).
The intensity of periurbanisation increased between the late 1960s and 1970s, as indicated by the generalised decline in net migration of dense and semidense areas. The declines were most marked (in precedence) among individuals aged 15−24, the children, the young working age population holding a postobligatory school diploma (i.e., secondary or tertiary level) and the mature working age population with a secondary education level. These groups increasingly favoured sparse and ring areas, thereby inflating net migration there. Between the late 1970s and 1980s, the behavioural changes F I G U R E 5 Continued LERCH | 9 of 14 were similar, except a reduced net migration among the working age population holding a postcompulsory education in the agglomeration rings.
During the shift towards re-urbanisation between the late 1980s and 1990s, net migration in dense urban zones increased among individuals aged 15−24, the young working age population holding a tertiary education level and, to a lesser extent, those with a secondary level. Correspondingly, the migration balance of these groups declined in the remaining agglomeration density zones.
In the most recent period of resumed periurbanisation (2010−2018), the situation switched again. When compared to the late 1990s, net migration among children and the mature working age population holding a postcompulsory school diploma decreased in dense and semidense areas. The decline in the dense areas' crude rate was intensified by lower net migration among individuals aged 15−24 and, to a lesser extent, the young and low qualified working age population. These negative contributions were only partly compensated for by an increasing migratory balance among the young and highly skilled working age population, as well as among pensioners.
Correspondingly, the children, individuals aged 15−24, the young and low skilled working age population, as well as the mature working age population with postcompulsory education, moved to a larger extent into sparse and ring areas, when compared to the late 1990s, thereby inflating net migration levels there. The positive contributions to the crude migration balance in sparse zones and agglomeration rings have been only marginally counterbalanced by a lower inflow of pensioners and the mature working age population with a secondary education level. In short, periurbanisation has recently superseded re-urbanisation mainly because the highly migratory youth and young adults head less frequently into dense areas, as well as because the out-migration of families to the urban periphery intensified again (the rising net migration of the individuals aged 15−24 in semidense zones did not compensate for the other groups' negative impacts on the corresponding crude rate).
Differences across the urban hierarchy in terms of the main groups responsible for the changes in crude net migration of urban density zones only exist with regard to the dense areas (not shown).
In the five largest agglomerations (Zurich, Basel, Geneva, Lausanne and Bern, with at least 300,000 inhabitants each in 2015), reurbanisation concerned the individuals aged 15−24 and the young and highly skilled working age population. In intermediate-sized cities (with populations between 100,000 and 300,000), however, the process was initially driven only by the 15−24 old age group and, since 2010, also by pensioners. In small cities, re-urbanisation resulted from higher net migration among the 15−24 old as well as F I G U R E 6 Contributions of behavioural changes in age-and education-specific groups to the evolution in the crude 5-year net migration rates by agglomeration density zone, Switzerland 1966.Sources: Population Censuses 1970, 1980, 1990, Population Register and Structural Survey 2015 among the young and mature working age population without tertiary education. However, the process was particularly short-lived there, when compared to larger cities. It has completely disappeared in the last observation period. In all those city-size classes, the renewed intensification of periurbanisation clearly outweighted the latest trend in re-urbanisation.

| SUMMARY AND DISCUSSION
Large-scale periurbanisation since the second half of the 20th century not only affected population health, notably by transforming the urban mortality gradient (Lerch et al., 2017), but also environ-  (Kulu & Steele, 2013).
While migration in the life stage dedicated to education is declining, spatial mobility increasingly matters in the subsequent professional life. The population in working ages moves more and more over time-whatever the educational attainment. With the additional diffusion of tertiary education in society, the highly skilled (and generally more migratory) subgroup is increasingly represented in the working age population, thereby inflating its crude rate of internal migration-as observed in Australia (Kalemba et al., 2021). In Yet re-urbanisation was a passing phase. Periurbanisation recently intensified again and increasingly focuses on the rural rings around agglomerations-rather than on the sparse urban zones closer to the centres. This may be explained by the soaring rental prices in Switzerland (which increased by at least 31% since 2000 according to the SFSO's rental price index), as observed in Germany (Osterhage, 2018;Sander, 2014;Stawarz et al., 2021). In Switzerland, gentrification particularly concerned the dense zones in large agglomerations (Rérat et al., 2010). Our results suggest that the LERCH | 11 of 14 15−24 old, the young and low-skilled working age population, as well as families, have been 'evicted' from dense zones to other more affordable residential areas in the urban periphery. This interpretation is plausible with regard to similar observations in other European countries (Dembski et al., 2021).
At the same time, the highly skilled working age population continues to afford a relocation to dense urban zones. Thus, agglomeration centres more and more constitute dynamic places of entertainment and elevator regions for young workers on an ascending professional track (Fielding, 1993): being in the hubs of the Swiss economy helps them better capitalising their skills to acquire the resources necessary to find a high quality residential A methodological note on the discrepancy between Rérat's (2012Rérat's ( , 2019 positive assessment of the Swiss trend in reurbanisation and our contradicting results is warranted. Our time series are not confounded by boundary changes and reclassification of municipalities over time (as we harmonised the administrative settlement structure). We rely on another (internationally comparable) classification of agglomeration areas, while accounting for an additional agglomeration zone (e.g., the rural ring). Rather than analysing trends in total population growth, as Rérat did, we have focused on the spatial redistribution of Switzerland's resident population to better understand how denser and sparser agglomeration zones are affected in a relational perspective. We have also neglected international migration flows (but not the internal movements of the immigrant stock). The historical peak levels and the highly-skilled profile of international inflows since the turn of the new century can be interpreted as yet another period effect that temporarily inflated the renewed population growth in dense zones.
More research is needed to better understand the role of international immigration and emigration alongside their interactions with internal movements in the process of city growth. Potential scenarios are multiple. A continuously large international inflow may be necessary to compensate for the immigrant stock's international re-emigration from Switzerland and its periurbanisation within the country.
Our results have implications for Switzerland's urban structure.
With renewed intensification of periurbanisation, the country is likely to become one single mega-urban region soon-rather than a set of interacting, densely populated but spatially distinct metropolitan areas.
Agglomeration ring areas will constitute urbanised interstices between formerly separated agglomerations and will become ever more attractive locations for dual-earner couples working in adjacent cities.
Despite recent official directives that aimed at limiting urban sprawl, many municipalities located in agglomeration ring areas have already consumed all the allotted constructible land for the next 20 years. If local authorities really want to reverse the current trend and limit negative externalities of urban sprawl on the environment and population health, they need to intervene more strongly in the zoning and housing market. Given the decentralised institutional structure of Switzerland, more cooperation between administrative units would help imposing urban containment measures, which are challenged by competition for tax payers. Moreover, to ensure inclusive and sustainable densification of cities, social mixing of the habitat is key.
The construction of social housing and housing cooperatives in central areas of agglomerations can avoid the eviction of individuals who aspire to live in the city and limit urban sprawl.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from Swiss Federal Statistical Office. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from the author(s) with the permission of Swiss Federal Statistical Offfice.