Beyond advancing a shared understanding of the concept, reviewing the defining features of urban shrinkage is valuable insofar as it serves to demonstrate what differentiates so-called ‘shrinking cities’ from other cities. Notwithstanding some key commonalities among existing efforts to define the phenomenon, there are several ways of understanding what constitutes urban shrinkage. While population decline is widely regarded as the central element, symptoms of shrinkage are as numerous as they are diverse. More precisely, urban shrinkage refers to a range of interconnected quantitative and qualitative changes occurring within a delimited urban area and over a specified period of time. Nonetheless, depopulation is widely regarded as the key quantitative indicator of shrinkage. Strictly speaking, a population shrinks either when the number of deaths exceeds births and this deficit cannot be offset by immigration, or, conversely, when net migration losses cannot be offset by natural population growth. However, the degree of depopulation that qualifies as urban shrinkage can vary. Under broad definitions, a shrinking city is any city where the peak population was recorded prior to the latest available figure, while under more stringent definitions, only those cities that meet a minimum threshold of population decline over a limited time period are classified as shrinking (Haase et al., 2014; Hartt, 2019; Wolff and Wiechmann, 2018).

Declining fertility is a frequent cause of excess natural population losses and quantitative decreases in population. Concurrent with increasing life expectancy, shrinking cities tend to exhibit faster rates of population ageing and higher concentrations of elderly residents than growing cities. In addition, shrinking cities typically experience increased rates of selective out-migration among higher-income and upwardly mobile households alongside increasing concentrations of socio-economically disadvantaged population groups (Galster, 2019; Martinez-Fernandez et al., 2016). The loss of jobs, particularly in the industrial or manufacturing sectors, is another common issue in shrinking cities (Wiechmann, 2008). Consequently, the average household income in these cities tends to decline over time (Hartt, 2019). Furthermore, especially in the context of western Europe, urban shrinkage often results in increases in vacant housing, industrial brownfields and otherwise un(der)utilised spaces. These demographic, economic and spatial changes can be caused or catalysed by critical events such as the introduction of disruptive political or economic reforms, environmental catastrophes, health crises or wars; but also by more gradual shifts in demographic dynamics, such as falling birth rates. Thus, natural population development can be both a cause and an effect of urban shrinkage (Martinez-Fernandez et al., 2016).

Beyond regional and global drivers, local governance arrangements as well as associated policy and planning strategies may influence the dynamics of urban shrinkage. However, in a neoliberal political climate that prioritises economic growth over social or environmental progress, shrinkage, along with its inherent fiscal pitfalls, may not be easily accepted by local decision-makers, let alone acknowledged (Bernt et al., 2014; Haase et al., 2014; Hospers, 2014). Shrinking cities that focus their efforts on bolstering economic development by means of conventional growth-oriented strategies, such as encouraging external investments or enacting tax breaks for the private sector, could ultimately exacerbate out-migration and socio-spatial inequalities, and thereby compromise their broader goals, like promoting social cohesion and life satisfaction (Fol, 2012; Maes et al., 2012). Alternatively, the conditions of shrinkage may present unique opportunities for cities to break free from contemporary trends of austerity urbanism and privatisation that tend to concentrate investments in prime areas at the expense of others. While emphasising the risks associated with top-down, centralised planning and triage measures for shrinking cities, many authors have argued that community self-determination and bottom-up planning are key to moving towards more holistic, sustainable and socially-equitable planning cultures (Berglund, 2020; Hollander and Németh, 2011; Pallagst et al., 2021).

Recognising the need to conceptualise urban shrinkage and to integrate it into broader theoretical debates in the fields of human geography, urban studies, urban and regional planning, and the social sciences at large, Haase et al. (2014) developed a heuristic model of urban shrinkage that incorporates the variety and interplay of the context-dependent mechanisms that underlie the phenomenon (Figure 1). While the model does not explain every case of urban shrinkage, it establishes a framework for facilitating more conceptually rigorous and context-sensitive empirical research on shrinkage, including on its causes, its consequences and the responses as well as the feedback loops and interrelations between these mechanisms.

By developing a conceptualisation of shrinkage independent of local or national characteristics, while also highlighting the reality of a diverse world of shrinkage, the heuristic can be used to conduct further (comparative) studies of shrinkage across various scales and geographies, thereby decoupling the phenomenon from its earlier, more straightforward definition as an invariant process unshaped by (supra)local contexts. The heuristic also illustrates how the interrelated processes of shrinkage may be sufficient to trigger a vicious cycle in which depopulation leads to both urban decline – in the form of increasing vacancies, loss of amenities and loss of attractiveness – and residential segregation, which may, in turn, fuel further depopulation.

Where the heuristic falls short is in illustrating that urban shrinkage does not necessarily involve the entire area or population of a city, but may instead affect certain areas and groups in particular. Due to persistent disinvestment and urban decline, parts of a city that continuously depopulate year after year, or even decade after decade, are likely to be worse off than areas that are only briefly affected by depopulation. Out-migration driven by deindustrialisation and other forms of economic change, for instance, mainly occurs among young adults and other relatively mobile segments of the population, including recent graduates and new families. By contrast, population losses in cities that shrink due to political turmoil or violent conflicts tend to be driven by the out-migration of young men. In cities where shrinkage is triggered by environmental catastrophes or health crises, it is the most socio-economically vulnerable population groups who are most likely to experience disproportionate declines.

Residential segregation refers to the differential distribution of social groups across (intra-urban) space. The literature has identified three quasi-universal types of residential segregation in contemporary cities:

demographic segregation: spatial concentrations of similar age groups or household types;

Broadly speaking, residential segregation results from at least two preconditions: heterogeneous population structures and differentiated housing market segments (Massey and Denton, 1988). Kovács (2020) described three additional factors that contribute significantly to residential segregation in cities: (1) income and wealth inequality, (2) the housing market and (3) ethno-religious differences. The influence of income inequality depends on the residents’ purchasing power, which basically determines who gets to reside in certain neighbourhoods. The tendency of the wealthiest members of society to spatially distance themselves from the lower classes by seeking out ‘higher-quality’ (read: larger, detached) housing in ‘nicer’ environments is evident in capitalist societies around the world. Poorly functioning housing markets and imbalances in the housing stock can also shape segregation. Cities that permit or encourage newly-built homes in suburban areas are effectively opening a window of opportunity for the better-off strata to self-segregate. Ethnic or religious differences may also have an influence on segregation patterns insofar as there is a general tendency among people with similar backgrounds to live near one another (Tammaru et al., 2020). Clark (1991) noted that the logic behind the formation of tightly-knit ethno-religious communities can be pragmatic, as wider social networks have been shown to provide individuals with a stronger sense of safety, as well as improved access to shared services and facilities. Lastly, the socio-cultural context of cities – or, more specifically, how the most mobile population groups seek to separate themselves from others – plays a decisive role in shaping past and contemporary socio-spatial differences. While many authors have observed that affluent people tend to congregate in the most attractive neighbourhoods in cities around the world, the dynamics and levels of segregation can differ greatly across various socio-cultural contexts (Kovács, 2020).

It therefore appears that residential segregation is a complex process, the dynamics and intensity of which hinge on a myriad of factors, ranging from geographical and institutional to socio-economic and spatial characteristics. In addition to migration flows, the attitudes, behaviours, culture and economic and political institutions of urban settlements seem to affect patterns of residential segregation. While such factors may adequately explain differing intensities of segregation among cities and countries, to understand the dynamics and patterns of intra-urban residential segregation – such as at the level of districts, neighbourhoods, blocks or even individual multi-household buildings – a more nuanced account of the role of context-specific causal mechanisms is needed. Accordingly, a context-sensitive perspective that thoroughly reflects on the history and the geography of cities, as well as on the relationships between formal and informal actors, is essential not only for understanding residential segregation, but also for developing context-specific measures to combat socio-spatial divides.

When examining shrinking cities, due consideration should be given to the effects of intermittent or continuous net migration losses and reduced housing demand. Moreover, it is important to recognise that conditions of shrinkage usually contribute to underinvestment in building and infrastructure maintenance, the proliferation of urban decay and relatively high levels of housing vacancy. Consequently, in shrinking cities, policies that aim to manage a surplus of low-cost housing are often prioritised over policies that seek to improve access to affordable housing in more sought-after neighbourhoods. As such conditions tend to cause property values and rents to stagnate or even fall, average incomes may decline as well. This is because more affluent and mobile households tend to comprise the bulk of out-migrants from shrinking cities, leaving behind larger shares of lower-income and minority households. Given that such population groups are generally less mobile, these dynamics can lead to growing concentrations of socio-economically vulnerable and otherwise marginalised households in the least attractive corners of a city (Großmann et al., 2013, 2015).

The former strand – that is, those studies that aim to unearth and describe the causal mechanisms behind changing socio-spatial distributions and divisions across one or more selected lower level administrative units, such as municipalities or communes characterised by urban population decline – centres on the observation that cities facing (systemic) urban shrinkage appear to be at an additional disadvantage in fostering social integration and cohesion compared to growing cities (Cortese et al., 2014). This is because some of the key drivers of shrinkage, including deindustrialisation, neoliberalisation and growing exposure to forces of globalisation, are often accompanied by increasing levels of unemployment, selective out-migration, accelerated ageing, housing vacancies and the erosion of public infrastructure and urban amenities. At the same time, shrinking cities tend to struggle with a general lack of capacity and resources to respond to such pressures, let alone to worsening socio-spatial inequalities. In sum, this strand posits that urban shrinkage, particularly of the long-term variety, effectively alters the socio-demographic make-up and spatial fabric of cities, and thereby threatens to act as a catalyst of residential segregation (Großmann et al., 2013).

The influential work of Glock and Häussermann (2004) reviewed the social consequences of shrinkage and, referring to prior empirical research on the case of Leipzig, described how neighbourhoods affected by depopulation face an increased risk of becoming trapped in a vicious cycle of physical and social downgrading. The authors explained that despite the apparent advantages of shrinkage for existing tenants, such as stagnating or decreasing rents and increased housing market options, high vacancy rates may have negative consequences for social cohesion over time, as reduced demand for housing and services can fuel a self-perpetuating process of rising costs and decreasing efficiency. In the late 2000s, Shrink Smart: The Governance of Shrinkage within a European Context, a research project funded by the European Commission under the Seventh Framework Programme (FP7) of the European Union, further fleshed out the groundwork for several important initial findings regarding shrinkage and socio-spatial inequalities, especially concerning changing distributions of socio-economically vulnerable or otherwise marginalised groups with restricted access to housing, such as low-income households, minorities and the elderly. These findings were synthesised in separate works by Großmann et al. (2013) and Cortese et al. (2014), both of which reviewed several post-hoc hypotheses regarding the impact of urban shrinkage on socio-spatial inequalities in light of the experiences of the cities of Genoa, Leipzig and Ostrava, a group of large, post-socialist cities challenged by deindustrialisation, labour market mismatch, employment migration, declining fertility rates and suburbanisation. Großmann et al. (2013) observed that although the national contexts and the trajectories of shrinkage of these three cities differed, depopulation was an additional explanatory factor in the emergence of urban decay and increased levels of socio-spatial segregation in all of them, largely because depopulation tended to be socially selective, and thus reinforced the marginalisation of neighbourhoods and concentrations of disadvantaged groups. Cortese et al. (2014) additionally analysed and compared policy efforts to promote social cohesion, and found that although policymakers in all three cities had some awareness of the increased risk of worsening socio-spatial inequalities under conditions of shrinkage, the public policies they introduced to promote social cohesion were neither comprehensive nor prioritised. Indeed, the authors observed that socially-oriented policies largely remained in the shadow of other, seemingly more important concerns, including economic revitalisation, housing market stabilisation and urban redesign and renewal. However, the authors also acknowledged that all three cities arguably lacked the capacities to effectively respond in the first place. Thus, it appears that social cohesion represents a specific challenge for shrinking cities.

In a similar vein, Fol (2012) reviewed past studies on socio-spatial change in several shrinking cities of France, and found that shrinkage, regardless of its cause(s), led to increasing socio-spatial disparities and the social exclusion of vulnerable social groups. In most of these shrinking cities, the cumulative effect of an exodus of better-off households, a depreciated real estate market and an increase in housing vacancies was to attract the most socio-economically vulnerable households to depopulating areas, especially those with increased availability of public housing or older, unrenovated apartments. For example, in medium-sized cities, such as in Roubaix and Saint-Denis, where shrinkage triggered by deindustrialisation had marked socio-spatial repercussions, the persistent out-migration of better-off households in search of an improved living environment resulted in increased concentrations of low-income and non-native groups in dilapidated buildings and large social housing estates in the city centre. Likewise, in smaller cities, such as Vierzon, which had in recent decades increasingly lagged behind its larger peers in terms of population growth, economic opportunities and information flows, low-income households were observed to concentrate in run-down parts of the old centre, as well as in peripherally-located social housing with an abundance of vacant apartments. Ultimately, it was the most vulnerable population groups who were most affected by shrinkage-related processes, such as rising unemployment and deteriorating living standards. Fol (2012) also noted that although municipalities and local actors may be best equipped to ensure socially-equitable outcomes for policy responses to the challenges of shrinkage, tensions between the pressure to respond to the needs of existing residents and the desire to attract new residents or businesses may fuel a worsening socio-spatial divide. Urban development policies promoted under the guise of social mixing may serve as a justification for planning strategies with an implicit aim of gentrifying certain areas of the city and attracting new middle-class residents that ultimately lead to the development of pockets of relative wealth, and thus exacerbate socio-spatial fragmentation. In the case of Saint-Étienne, the local discourse surrounding social mixing served as a justification for the implementation of ‘soft gentrification’ tactics in the city centre, which included new housing developments, public space redesign projects and a variety of cultural offerings. By drawing better-off residents to the area, these tactics initially worked to curb concentrations of socio-economically vulnerable groups in the city centre. However, the ensuing rise in property values sowed the seeds of new challenges for existing lower-income residents, ranging from rent gouging to involuntary displacement. Meanwhile, in Saint-Denis, increased concentrations of unemployed persons and accompanying social problems were observed in areas of the city that were excluded from a large-scale urban revitalisation project that aimed to both demolish the city’s most degraded housing stock and upgrade its most valued buildings. Hence, Fol (2012) suggested that policy interventions aimed at improving the living conditions of marginalised areas and increasing housing choices for low-income households may be more effective at combating increasing socio-spatial disparities such as segregation than social mixing strategies based on rightsizing or economic development. The author concluded by stating that while increasing socio-spatial disparities seem to be a common outcome of the processes of and the responses to shrinkage, further research distinguishing the spatial scales at which socio-spatial fragmentation occurs is necessary.

Towards that aim, Maes et al. (2012) considered the socio-spatial effects of shrinkage in the depopulating working-class neighbourhood of Hrušov in Ostrava, Czech Republic, arguing that in order to understand the effects of shrinkage on everyday life, it is essential to seriously consider the agency of individuals and households. Based on interviews with existing residents, the authors concluded that while shrinkage undoubtedly had negative effects on everyday life in certain neighbourhoods, these effects were especially severe in declining areas of post-socialist cities like Ostrava, where the transition to capitalism (and the widespread privatisation of housing) was an added obstacle to the development of socially-mixed neighbourhoods, and where instances of ‘planned shrinkage’ markedly altered the ways in which inhabitants lived. In Hrušov, the combined effect of institutional change and shrinkage led to concentrations of socio-economically vulnerable minorities, particularly the Roma. Here, the post-socialist transition led not only to a reduction in affordable housing, but also to the near complete withdrawal of social infrastructure and leisure facilities. The authors concluded that conditions of shrinkage and accompanying resource constraints represent considerable obstacles to socially-equitable urban development and the pursuit of social justice. Maes et al. (2012) observed that social mixing and social cohesion are especially challenging goals under conditions of urban shrinkage, given the financial and structural deficits it generally entails; in other words, that shrinking cities tend to lack the resources and capacities necessary to combat socio-spatial inequalities.

In another key work considering context-specific paths of socio-spatial change, Haase et al. (2012) coined the term ‘transitory urbanites’ to describe a new type of highly-mobile resident in formerly depopulating inner-city neighbourhoods of second-tier cities in Czechia and Poland. Specifically, the authors found that an increasing share of younger households – many of whom emotionally and pragmatically valued inner-city living in their current phase of life, but would consider relocating to peripherally-located neighbourhoods or even suburbia in the short to medium term – relocated to these areas beginning in the second decade of the region’s post-socialist transition. The authors noted that this socio-spatial shift would not have been possible without the highly-fragmented inner-city housing market structures specific to many post-socialist cities that emerged due to urban shrinkage and the dilapidation of older built-up neighbourhoods. Essentially, many young newcomers managed to take advantage of the particular housing market conditions by renting a newly-privatised apartment, be it from the legal tenant, a private housing company or a relative. Although their tenures were often temporary, the behaviours and residential preferences of these transitory urbanites nevertheless paved the way for future influxes newcomers of similar ages and social status to inner cities. Thus, these temporary residents may have contributed to processes of studentification or gentrification (Haase et al., 2012).

The latter strand of case study research focused on patterns of residential segregation in shrinking cities includes studies that apply descriptive statistical techniques to measure concentrations of marginalised population groups at the level of districts or neighbourhoods, and thereby help to identify which intra-urban areas are the least and the most exposed to socio-spatial inequalities. While the first of these studies employed more traditional descriptive measurements such as averages, shares, frequency tables or standard deviations, recent studies have used more complex datasets, more advanced statistical techniques, more granular qualitative analyses and more intuitive methods of data visualisation. The ground-breaking work of Petsimeris (1998) used the indices of segregation and dissimilarity as well as location quotients to examine the changing spatial arrangement of social and ethnic population groups in the metropolitan areas of Milan, Turin and Genoa: i.e., three cities that form the so-called ‘industrial triangle’ of Italy. Four types of residential areas shared by all three cities were distinguished: the historic centre, a peri-central belt, upper-class quarters and the outer city. Using census data on occupational status from 1981 and 1991, which represents a period at the tail end of a wave of widespread deindustrialisation and deurbanisation in Italy, a positive effect was observed between inner-city population decline and increasing levels of residential segregation, above all with regard to concentrations of working-class groups in peripherally-located neighbourhoods near industrial zones, but also to concentrations of professionals and business owners in historic city centres. The author attributed this effect to shifts in the absolute and relative shares of lower and higher socio-economic groups, as well as to generally higher levels of out-migration and intra-urban residential mobility among affluent households in all three cities.

Subsequently, Großmann et al. (2013) observed that conditions of urban shrinkage contributed to the formation of niches of lower-income households in areas of low-cost housing characterised by abandonment, vacancies and decay. The outflow of residents reduced public and private investment, which, in turn, exacerbated these problems. However, the affected social groups and patterns of socio-spatial differentiation differed across the cities in question: in Ostrava, the Roma population were increasingly excluded from the more desirable parts of the city; in Genoa, concentrations of low-income households increased in the shrunken areas of the historic city centre; and in Leipzig, the shares of unemployed persons and ethnic minorities living in older working-class neighbourhoods increased. These findings demonstrated that the impact of shrinkage on residential segregation was context- and path-dependent. Additionally, in Ostrava, small-scale instances of demographic and ethnic residential segregation were observed, with older, long-time residents and younger newcomers increasingly residing on different streets and in different buildings. In Genoa, a sort of vertical residential segregation was observed in the shrinking inner-city quarters that was characterised by higher-than-average concentrations of better-off households living on the upper floors of otherwise vacant buildings. Similar observations were made in the district of Leipzig-Grünau, a large housing estate where depopulation alongside selective demolition and upgrading measures contributed to growing income disparities between neighbouring apartment buildings. However, these spatial divisions were so fine-grained that they remained undetected when calculating segregation indices at the level of districts or even neighbourhoods. Großmann et al. (2013) concluded that although the link between shrinkage and segregation appears to largely depend on the local context and the drivers and the scope of shrinkage, as well as on the scale at which the phenomena are analysed, there is strong evidence that urban shrinkage has a catalysing effect on the (rapid) emergence of pockets of poverty and affluence within specific types of residential areas.

Großmann et al. (2015) further investigated the influence of housing oversupply on socio-spatial change and differentiation in Leipzig; a case that provided the researchers with the rare chance to investigate the impact of a city-wide housing surplus on the dynamics and the patterns of residential segregation, as opposed to a glut of vacancies confined to a specific area. The authors found that as the city’s population declined during the 1990s, levels of residential mobility peaked and certain districts experienced a thorough reshuffling of residents. Using several different index-based measurements of segregation covering the five dimensions of residential segregation described by Massey and Denton (1988): evenness, exposure, concentration, centralisation and clustering, the authors found that Leipzig’s city-wide housing oversupply and falling property values opened up niches for socio-economically vulnerable and minority population groups, above all in substandard housing in less attractive neighbourhoods. The authors also utilised GIS to illustrate how unemployed and non-native residents became increasingly concentrated in unmodernised, large, peripherally-located housing estates and in select inner-city neighbourhoods characterised by decay, respectively. An analysis of the average ages of the residents showed that there was an influx of young people into inner-city districts, while outer core and suburban areas grew older. Given that spatial distributions of depopulation and vacancies strongly correlated with concentrations of vulnerable groups, it appears that urban shrinkage contributed to Leipzig’s socio-spatial restructuring, and to the partial re-emergence of the city’s pre-war patterns of residential segregation.

Slach et al. (2019) examined how urban shrinkage impacted social sustainability in Ostrava, a sprawling city that experienced a decline in population between 1991 and 2018 due to post-socialist economic reforms, deindustrialisation, out-migration, suburbanisation, environmental degradation, declining fertility rates and a reliance on pro-growth policy strategies focused on securing financial investments or fostering economic growth. The authors concluded that shrinkage deepened existing patterns of residential segregation, which were characterised by concentrations of higher- and lower-status households in more and less desirable inner-city locations, respectively. The areas with the most pronounced levels of social exclusion were mainly concentrated in inner-city neighbourhoods characterised by higher levels of abandonment or in more peripherally-located areas near brownfields. Interestingly, while cities in Central and Eastern Europe (CEE) like Ostrava have been particularly prone to shrinkage since the turn of the millennium due to the region’s post-socialist political and economic transition, unlike many shrinking cities in western Europe and America, most have not faced any major challenges related to housing vacancies, given the region’s longstanding and widespread housing shortages dating from the era of state socialism. Accordingly, shrinking cities in CEE have generally not experienced the housing market imbalances or the demolition efforts that have occurred in the shrinking cities in the West. Nonetheless, the role of governmental and market actors in shaping socio-spatial change through public policies, urban planning and investment in Ostrava, as well as in many other shrinking cities, should not be ignored. It appears that at least during the initial years of Ostrava’s post-socialist transition, governments saw unregulated markets as a mechanism of resource allocation that would facilitate prosperous, economically-efficient and socially-just societies. Since shrinkage was widely considered a temporary phenomenon stemming from the post-socialist transformation, the prevailing policy strategy remained focused on attracting external investment, with decision-makers assuming that economic growth would eventually translate into population growth.

Hoekstra et al. (2020) assessed the socio-spatial consequences of housing policies introduced to respond to depopulation in Parkstad Limburg, a region with approximately 250,000 inhabitants, including approximately 90,000 in the municipality of Heerlen. The analysis focused on changing patterns of socio-economic segregation and housing accessibility among lower-income households between 2004 and 2015. The authors drew on secondary statistics to track changes in the region’s socio-economic composition, as well as to map its geography of population decline and growth. Additionally, they analysed levels and patterns of residential segregation across the region, employing both the index of dissimilarity and the Theil index. The latter offers two noteworthy advantages over the former: first, the Theil index allows for the measurement of spatial evenness both between and within intra-urban units; and, second, it utilises continuous data (in this case on income), and thus does not have to rely on categorical variables such as occupational status. Even though shrinkage is less common and less severe in the Netherlands than elsewhere in Europe and the United States, the authors posited that the case of Limburg nevertheless offers a unique opportunity to investigate the effectiveness of policies specifically designed to mitigate population decline and to balance a (highly-regulated) housing market. In order to analyse the effects of housing policy interventions on population distributions, Hoekstra et al. (2020) mapped the geography of housing construction and demolition over time, and examined how these efforts served different income groups, and thereby altered levels and patterns of segregation. The authors found that even though a decree was issued stating that new construction had to be compensated for by the demolition of existing properties, and efforts were made to distribute demolitions and new developments evenly across the region, the construction of new housing in promising locations tended to go hand-in-hand with the demolition of low-quality housing in less desirable areas. Combined with growing economic polarisation and the gradual ageing of the region’s population, these policies appear to have contributed to a further deepening of existing socio-spatial divides. These occurred as low-income households became substantially overrepresented in high-density areas, while the already large shares of better-off households in more sprawling areas became even larger relative to the shares of low-income households. Simply put, a widening spatial gap between poor households in inner-city neighbourhoods and relatively affluent households in suburban and rural areas was observed in the wake of shrinkage. Policy responses to depopulation had mixed effects on these dynamics; however, a tendency to focus efforts on the selective demolition of relatively inexpensive housing in high-density areas, while new housing construction mainly catered to middle-class households and was more evenly distributed across the region, disproportionately impacted low-income residents.

In order to illustrate the role of urban shrinkage in dynamics and patterns of residential segregation as described in the literature, Figure 2 presents a high-level overview of how various facets of these phenomena interlink with each other. For urban shrinkage, typical symptoms that have been identified as drivers of segregation are highlighted. Similarly, for residential segregation, a selection of symptoms that are particularly relevant from the perspective of shrinkage are listed, including city-wide socio-spatial unevenness, intra-urban spatial concentrations or clustering of certain population groups, as well as other patterns of intra-urban socio-spatial differentiation or inequalities. The diagram shows symptoms of shrinkage that have been identified as potential causal mechanisms of segregation. In addition, it indicates how local governance structures or urban planning responses to these symptoms may result in spatially-just or unjust outcomes (Berglund, 2020; Fol, 2012; Hoekstra et al., 2020).

Essentially, the socially-selective population losses and the altered housing market conditions typical of shrinking cities tend to lead to urban disinvestment and decline and to increasing perforation of the urban fabric, which may, in turn, trigger shifts in the city-wide extent or intra-urban patterns of residential segregation. Such shifts are especially likely to occur when there is a sudden, shock-like driver of shrinkage, such as a major political or economic change, the collapse of an industry, or even a natural catastrophe. Such events have been observed to result in both rapid and unexpected socio-spatial change (Großmann et al., 2015; Petsimeris, 1998). Thus, the dynamics of spatial inequalities and segregation under conditions of shrinkage differ from those typically observed under conditions of growth and ongoing housing shortages. Evidently, shrinkage can result in growing concentrations of disadvantaged households in the least attractive neighbourhoods with an oversupply of relatively cheap housing stock, in addition to a lack of public infrastructure or urban amenities. Continuing selective out- and in-migration alongside efforts to reuse or modernise certain properties or spaces may result in small-scale instances of segregation, such as at the level of neighbourhood blocks or even individual buildings. However, the longevity of such developments – and whether these places eventually change due to, for example, reurbanisation or gentrification – depends largely on the local context, including on its housing market characteristics, future population trajectories, and urban policy and planning priorities. While context-sensitive policy and planning responses to shrinkage and segregation can facilitate equitable and just outcomes, the experiences of numerous cities indicate that non-strategic responses could make matters worse (Berglund, 2020; Fol, 2012; Pallagst et al., 2021).

Previous studies have improved our understanding of the general dynamics as well as the context-specific paths of residential segregation in a variety of shrinking cities around the world. At the same time, these studies raise a number of as yet unanswered questions about the causal nexus and the mechanisms at work between urban shrinkage and residential segregation. While their findings largely align with the theory that urban shrinkage, however defined, may act as a catalyst for residential segregation by age, ethnicity or socio-economic status, the scope of empirical evidence remains rather limited, and is predominantly based on the experiences of a handful of large capitals and metropolises. Indeed, with few exceptions (e.g., Fol, 2012; Hoekstra et al., 2020), the evidence for causal relationships between shrinkage and segregation in small- to medium-sized cities is largely anecdotal. Furthermore, the bulk of empirical studies, and thus the extent of our knowledge, is based on the European context, where cities have traditionally been characterised by lower levels of ethnic and socio-economic residential segregation than is the case in, for example, the United States (Cassiers and Kesteloot, 2012). Thus, while urban shrinkage has become an increasingly widespread global phenomenon from a geographical perspective, empirical studies verifying that residential segregation worsened during or in the wake of shrinkage are few and, spatially speaking, far between. Consequently, there are significant blind spots in our knowledge of many regions around the world, and it therefore seems hazardous to make broad generalisations about the influence of urban shrinkage on dynamics or patterns of segregation.

There is also a need for more granular spatial analyses of patterns of residential segregation under conditions of shrinkage. Whereas regional and urban scale studies may help to identify local causes, trajectories or effects of shrinkage, intra-urban scale analyses allow for the identification and comparison of individual neighbourhoods, and are thus important for the development of a more granular understanding of the interplay between shrinkage and segregation. Additionally, as most past studies examined levels or patterns of segregation at a single point in time, longer-term perspectives on how shrinkage impacts segregation – and findings related to potential time-lag effects – are missing. While there may be little reason to doubt that urban shrinkage implies some degree of socio-spatial change, it is difficult to determine precisely to what degree or even whether the relationship is linear, exponential or neither based on the existing literature, which has, to be clear, centred on a few exemplary cases of shrinkage. To fill these gaps, researchers have increasingly called for more systematic comparative research of the socio-spatial effects of processes of and responses to shrinkage.

Although they are evolving, the statistical techniques employed by existing studies also leave some room for improvement. An important shortcoming of these studies is that they used different methods and indicators. While the studies reviewed above made much of the spatial context, because of data limitations, most focused on patterns of spatial evenness, and essentially brushed other dimensions of residential segregation – namely, exposure, concentration, centralisation and clustering – under the rug. In addition to using different statistical techniques to measure levels or patterns of spatial evenness, the few studies that explicitly analysed intra-urban patterns of segregation employed different spatial unit types and sizes. While some analyses were performed at the level of neighbourhoods or districts, others were conducted at a larger scale, such as boroughs or census tracts. Moreover, although the studies were predominantly concerned with spatial divisions along socio-economic lines, their definitions of disadvantaged and affluent groups differed. Aside from analysing education, income and employment status, these studies failed to consider other factors of potential interest, including levels of social and human capital.

There seems to be a lack of engagement with more critical approaches, such as the political economy perspective, and of studies that explicitly consider and compare context-specific pathways, dynamics and outcomes. In addition to a need for research on similarities and differences in levels and patterns of segregation among various types of shrinking cities around the world, little attention has been given to the question of how governance frameworks and actor constellations shape the agency of decision-makers to respond to the socio-spatial challenges presented by shrinkage, especially with regard to housing vacancies. Lastly, the socio-spatial effects of place-based policy responses to shrinkage – including frequently endorsed ‘smart shrinkage’ strategies, such as rightsizing, greening and vacant property reuse – remain opaque. Future research could, for instance, investigate whether urban development scenarios such as uncontrolled (sub)urban sprawl or top-down demolition programs have resulted in new challenges for marginalised neighbourhoods that impede their sustainable development and hinder potential improvements in the quality of life for existing residents. Such research might have global relevance by providing lessons in the unintended consequences of implementing ‘quick fixes’ in response to fundamental change. In sum, the methods and the empirical results of past studies on socio-spatial change and residential segregation in shrinking cities do not lend themselves to future post-hoc, cross-sectional or comparative research. While these studies together represent a solid first cut at investigating the question of how urban population decline affects segregation, several promising angles of inquiry remain to be pursued.